Fullständig transkription av This Dum Week 2026-10-25

but I learned— It used to 69 00:04:22,436 --> 00:04:23,643 stand for emergency. Now it stands for eventual. 70 00:04:25,986 --> 00:04:30,039 Give me an over and under. How long you think it was before 71 00:04:30,154 --> 00:04:32,878 I got seen? Three hours. 72 00:04:35,174 --> 00:04:36,754 That oh is that your over or under? 73 00:04:38,706 --> 00:04:40,934 Um, it's my it's my break my break even. 74 00:04:42,982 --> 00:04:45,904 Nine hours, Alex. Nine hours. 75 00:04:48,374 --> 00:04:51,765 Maybe they felt that since it since it happened at night. 76 00:04:52,486 --> 00:04:55,739 You should see an appropriate doctor. 77 00:04:56,492 --> 00:04:59,664 By 2 AM, Alex, by 2 AM, 78 00:05:00,113 --> 00:05:03,349 I was one of 4 people left in that emergency room waiting room, 79 00:05:03,878 --> 00:05:07,290 and I am curled over on a table rocking 80 00:05:07,386 --> 00:05:10,911 myself back and forth in partial delirium because not 81 00:05:11,411 --> 00:05:15,734 only had I not really slept, you know, I had that 5-hour period, but that 82 00:05:16,234 --> 00:05:19,980 was across a 2.5-day period. I needed more sleep than that, but I also 83 00:05:20,044 --> 00:05:23,539 hadn't had my caffeine In something like 48 hours. 84 00:05:23,587 --> 00:05:27,598 So now I'm going through caffeine withdrawal, I have— I 85 00:05:28,098 --> 00:05:31,898 had no sleep, and I'm writhing in agony due to extreme stomach pain, 86 00:05:31,930 --> 00:05:35,925 and nobody, nobody cared. I'm one of 4 people left. Finally, at 2 87 00:05:36,101 --> 00:05:39,438 AM, 2 AM, they, they checked me back in to see me, 88 00:05:39,454 --> 00:05:42,423 Alex. They stick me full of stuff, draw blood, do all that other stuff, 89 00:05:43,144 --> 00:05:46,786 and, uh, at 6 AM they say, okay, you can go home now. 90 00:05:47,958 --> 00:05:51,120 Uh, you'll— here's— you just take some over-the-counter pain meds. 91 00:05:51,185 --> 00:05:54,347 And that's, that's the best she got. And I'm like, well, okay, but listen, 92 00:05:54,941 --> 00:05:58,055 um, I can't spend 9 hours in the ER again. This was, 93 00:05:58,088 --> 00:06:01,250 this was torture. You don't understand. This was absolute hell. 94 00:06:01,378 --> 00:06:04,974 I can't do this, uh, again. So what, what is this? 95 00:06:05,135 --> 00:06:08,586 They're like, well, it might be viral. We don't think it's gall— your, 96 00:06:08,602 --> 00:06:11,749 like, gallbladder or something, because when we would have— when we, like, you know, 97 00:06:11,797 --> 00:06:14,799 inspected your abdomen, you would have screamed at us in, you know, 98 00:06:15,008 --> 00:06:17,929 blood-curdling fashion, and you didn't. And it's like, Okay, 99 00:06:19,019 --> 00:06:22,706 so we think it might be viral and just, you know, come back later 100 00:06:22,738 --> 00:06:25,880 if you can't actually keep any food down and you're puking everything 101 00:06:25,992 --> 00:06:28,669 out. And it's like, oh, okay, 102 00:06:29,759 --> 00:06:33,591 uh, Alex. Now, I, I, the, the good news is that the, the amount of 103 00:06:33,767 --> 00:06:37,166 over-the-counter pain meds they've authorized me to take seems to be enough 104 00:06:37,230 --> 00:06:40,933 to keep it in a manageable, uh, manageable stance. 105 00:06:40,997 --> 00:06:44,380 But this is— if there is any reduction, Alex, 106 00:06:44,444 --> 00:06:47,618 in the quality of the show today, um, Yeah, it's really— it is due 107 00:06:47,730 --> 00:06:51,419 to my General several days of 108 00:06:51,919 --> 00:06:55,237 misery with this experience. 9 hours, Alex. I'm telling you, that was, 109 00:06:55,413 --> 00:06:59,503 that was very bad. That was very bad. What happens though, Gator? We have 110 00:07:00,003 --> 00:07:03,498 to consider all alternatives as properly scientific-minded Americans. 111 00:07:04,251 --> 00:07:07,604 Okay. All right. What happens if there's an improvement in 112 00:07:08,104 --> 00:07:11,101 the show? What do we do then? Well, 113 00:07:11,614 --> 00:07:15,143 then we can finally, we can finally ascend to the 114 00:07:15,643 --> 00:07:19,025 true realm of artistry because artists are the ones who thrive 115 00:07:19,525 --> 00:07:23,071 in their own suffering. Um, if you, if you take away the suffering from 116 00:07:23,103 --> 00:07:25,785 the artist, then they have nothing. So if this show is better, 117 00:07:26,620 --> 00:07:29,832 then, then we finally have our first— I'm gonna have to, uh, show up at 118 00:07:30,332 --> 00:07:33,623 your place and slip you— slip some drugs in your drink. 119 00:07:33,703 --> 00:07:37,140 Dude, man. For the art! 120 00:07:37,317 --> 00:07:40,657 For the art! It was bad, man. I was, I was 121 00:07:40,834 --> 00:07:43,179 really— I was in dire straits, dude. 122 00:07:43,821 --> 00:07:47,756 It was not pleasant. So— Oh man, I feel you. I feel you. 123 00:07:47,981 --> 00:07:49,057 This, this is the worst. 124 00:07:51,206 --> 00:07:55,487 Having sampled healthcare in multiple continents, 125 00:07:55,695 --> 00:07:59,111 it doesn't get better. That's all 126 00:07:59,127 --> 00:08:01,019 I can tell you. Now, 127 00:08:02,189 --> 00:08:06,230 I would say my biggest complaint, Alex, my biggest complaint is when I went to 128 00:08:06,839 --> 00:08:10,238 process myself, go in and sign in, there wasn't a 129 00:08:10,270 --> 00:08:12,932 single human being who said, oh, by the way, 130 00:08:13,638 --> 00:08:16,652 we're having extraordinarily long wait times today. 131 00:08:17,662 --> 00:08:21,236 We do not know how long it's gonna take. It could take 9 132 00:08:21,252 --> 00:08:25,002 hours or more. Had someone told me that, I would've 133 00:08:25,018 --> 00:08:28,623 driven to other nearby ERs that possibly had 134 00:08:29,123 --> 00:08:32,197 different wait times. But nowhere 135 00:08:32,325 --> 00:08:35,979 did they tell a single human being that there was something wrong. It's only 136 00:08:36,075 --> 00:08:38,671 after about 5 hours that when I went up to

the counter and said, 137 00:08:38,751 --> 00:08:41,476 is there any indication you could give me about when I'm gonna be seen? 138 00:08:41,524 --> 00:08:44,953 They're like, no, we don't, we just, we don't know. This is just, 139 00:08:45,001 --> 00:08:48,719 it's so weird. Today is the, it's the worst we've ever seen. 140 00:08:49,797 --> 00:08:53,228 Like, okay. What are the odds? What are the odds? 141 00:08:53,808 --> 00:08:57,949 Okay, but, but I, I'm— I, 142 00:08:58,408 --> 00:09:01,148 I'm— I need help. 143 00:09:03,601 --> 00:09:06,506 So anyway, moving forward, Alex, 144 00:09:06,893 --> 00:09:10,266 uh, in our, in our miscellaneous stories that we use to begin the show, 145 00:09:11,449 --> 00:09:15,246 uh, serendipitously we have some medical malpractice. 146 00:09:17,690 --> 00:09:20,901 Actress Christina Applegate revealing doctors misdiagnosed her 147 00:09:21,401 --> 00:09:24,835 with MS. What they told her and what is the explanation now for 148 00:09:25,335 --> 00:09:29,395 her devastating symptoms? Here's Kayna Whitworth. For 5 years, 149 00:09:29,572 --> 00:09:32,783 Christina Applegate has candidly shared her battle against multiple 150 00:09:33,283 --> 00:09:36,059 sclerosis, becoming a voice for so many with the disease. 151 00:09:36,637 --> 00:09:39,880 But tonight, the actress revealing in an emotional podcast she 152 00:09:39,928 --> 00:09:44,035 was shocked when a doctor recently told Her she had been misdiagnosed. 153 00:09:44,676 --> 00:09:48,842 It's hard to say that she came into my room and she 154 00:09:49,342 --> 00:09:52,656 said, um, congratulations, you don't have MS. 155 00:09:53,473 --> 00:09:57,095 Applegate telling friend and co-host Jamie Lynn Sigler, who has MS 156 00:09:57,595 --> 00:10:01,630 herself, the news has been confusing. She says her symptoms remain. 157 00:10:01,839 --> 00:10:06,294 She needs a wheelchair to even go to the doctor. You don't get misdiagnosed 158 00:10:06,794 --> 00:10:09,888 and then all of a sudden I'm better. Right. I haven't gotten better, Jamie. 159 00:10:10,306 --> 00:10:13,662 Yeah. If anything, I've gotten worse in some 160 00:10:13,774 --> 00:10:17,131 ways. And so 161 00:10:17,388 --> 00:10:20,921 I'm back at square one. MS causes the immune system to 162 00:10:21,421 --> 00:10:25,225 attack cells in the central nervous system, and it can mimic other inflammatory diseases. 163 00:10:25,562 --> 00:10:29,159 Experts say up to 20% of cases can be misdiagnosed. 164 00:10:29,432 --> 00:10:32,821 MS is a difficult condition to diagnose. We don't have 165 00:10:32,949 --> 00:10:36,466 a single diagnostic test that can tell us that a person 166 00:10:36,482 --> 00:10:39,820 has MS. Darling, I did the news and I nailed it. 167 00:10:40,062 --> 00:10:43,959 The former star of Anchorman and Dead to Me telling our Robin Roberts 168 00:10:44,459 --> 00:10:48,257 earlier this year she was still struggling to accept the diagnosis of MS. 169 00:10:48,611 --> 00:10:52,037 Where are

you now with the grieving process of MS? 170 00:10:52,793 --> 00:10:53,790 Still sucks, Robin. 171 00:10:56,485 --> 00:10:59,618 It really does. And anyone that tells you different, 172 00:10:59,923 --> 00:11:03,168 they're not telling the truth. I'm sorry. Tonight, 173 00:11:03,297 --> 00:11:06,725 Applegate says she and her doctor are starting from scratch. And David, again tonight, 174 00:11:06,774 --> 00:11:10,121 Applegate says she's back Back to square one. 175 00:11:10,234 --> 00:11:13,964 She's working with a neurologist to find a new diagnosis, but an expert 176 00:11:14,029 --> 00:11:17,936 on her podcast said she's not alone in being misdiagnosed 177 00:11:17,985 --> 00:11:19,408 with MS. David. 178 00:11:21,770 --> 00:11:24,088 So that's a, that's a very sad story, Alex. 179 00:11:24,764 --> 00:11:27,485 Uh, I can't imagine what it's like to think for 5 years you have, 180 00:11:28,064 --> 00:11:31,236 uh, MS and then find out you, you don't and have no answers. 181 00:11:31,300 --> 00:11:34,608 That's pretty terrible. So I feel bad for Miss Applegate there. 182 00:11:36,841 --> 00:11:39,117 She can rename her podcast MS Previously. 183 00:11:45,250 --> 00:11:47,982 Too soon, Alex. Too soon. Too soon. 184 00:11:48,320 --> 00:11:51,518 Yes. Right now. Too soon. 185 00:11:54,010 --> 00:11:57,566 Look, it's just a use it or lose it joke. Okay. I guess, I guess 186 00:11:58,066 --> 00:12:01,009 it is. When would you have another opportunity? Yeah. Well, 187 00:12:01,300 --> 00:12:04,183 Well, let's see what you have for this next story, Alex, here. 188 00:12:04,509 --> 00:12:06,740 Well, but I have to say, 189 00:12:08,470 --> 00:12:11,353 first of all, so they don't have a diagnostic for MS. 190 00:12:11,547 --> 00:12:16,911 And my understanding of MS is that they don't even really know what 191 00:12:16,943 --> 00:12:19,097 it kind of truly is. 192 00:12:20,335 --> 00:12:24,208 It's a combination of symptoms that they sort of have some ideas 193 00:12:24,708 --> 00:12:28,234 about or whatever. You could say it's a mess. But what 194 00:12:28,282 --> 00:12:31,158 I don't understand is, like, 195 00:12:31,351 --> 00:12:35,415 what is this definitive test that they did that told them 196 00:12:35,447 --> 00:12:38,821 that she doesn't have the thing that they don't have a test for, but she 197 00:12:39,321 --> 00:12:42,517 has something else that they probably don't have a test for? Like, do you see 198 00:12:43,017 --> 00:12:45,890 that there's, like, some gaps in the story somehow? Yeah, yeah, yeah, 199 00:12:46,115 --> 00:12:49,762 that's a good question. What is it that disconfirmed the diagnosis 200 00:12:50,710 --> 00:12:53,313 that wasn't able to be caught? That they couldn't have confirmed. 201 00:12:53,917 --> 00:12:56,156 Yeah, yeah, yeah. That's a good question. 202 00:12:57,950 --> 00:13:01,038 So I, by the way, I can probably tell you what 203 00:13:01,538 --> 00:13:05,735 the comment section of those videos looks like, um, because I, I watched 204 00:13:06,273 --> 00:13:08,430 another video, similar vibes, 205 00:13:09,111 --> 00:13:12,679 from, uh, Science Girls. I don't know if you remember her. She's a YouTuber 206 00:13:13,179 --> 00:13:16,892 that was around a few years ago, uh, and around about 2021. 207 00:13:20,090 --> 00:13:24,086 She got COVID, and then she got better, and then she 208 00:13:25,429 --> 00:13:27,597 had this huge sort of exhaustion, 209 00:13:28,693 --> 00:13:31,094 like massive. Went down a long COVID track. 210 00:13:33,218 --> 00:13:35,678 Apparently, she has long COVID. 211 00:13:35,905 --> 00:13:39,368 But you know, again, you can imagine what the 212 00:13:39,707 --> 00:13:43,356 comment section looks like. Yeah. We've been known to be spicy 213 00:13:43,389 --> 00:13:46,371 with our COVID opinions. So. 214 00:13:47,350 --> 00:13:50,890 Um, I don't know. I want to see some dates. Oh, you said 5 215 00:13:50,906 --> 00:13:53,915 years, right? So since 2021 then? Yeah. 216 00:13:54,447 --> 00:13:57,103 Oh, okay. 217 00:13:58,391 --> 00:14:01,675 Oh, oh, I see where you're heading. I see where you're heading, 218 00:14:01,772 --> 00:14:04,608 sir. I, I don't know. That's what— those are good questions though. 219 00:14:05,401 --> 00:14:08,006 Now let's see, see what you have for this. Was it sudden onset? 220 00:14:08,264 --> 00:14:11,856 Non-MS? Sorry. I— yeah, 221 00:14:11,872 --> 00:14:14,991 I don't know, man. Yeah, but, 222 00:14:15,119 --> 00:14:18,534 uh, okay. Deadly Utah car crash. 223 00:14:18,806 --> 00:14:22,397 Cops say woman was looking at porn while driving. 224 00:14:23,166 --> 00:14:26,565 A woman caused a fatal car crash in Utah because she was too busy watching 225 00:14:26,597 --> 00:14:31,085 porn while driving, according to authorities. Multiple outlets reported the 44-year-old 226 00:14:31,585 --> 00:14:35,413 California woman Christine Chow was charged Tuesday with automobile homicide 227 00:14:35,913 --> 00:14:39,549 in connection with the deadly June 12th crash. According to charging documents, Chow told 228 00:14:39,565 --> 00:14:42,947 officers she was at fault, and when investigators got a warrant to search 229 00:14:43,447 --> 00:14:46,795 her phone, they found she'd accessed porn sights less than 2 minutes 230 00:14:47,295 --> 00:14:51,251 before the crash. The charging docs say Chow was driving on I-15 231 00:14:51,524 --> 00:14:54,906 when she crashed into a white sedan which had slowed down in a construction zone. 232 00:14:54,954 --> 00:14:58,481 The sedan was hit, then hit the SUV driving in 233 00:14:58,981 --> 00:15:02,841 front of it. All 3 vehicles went off the road. KSL reports the 66-year-old 234 00:15:02,889 --> 00:15:06,448 Utah man Scott LeVar Litchfield, who was riding in the backseat 235 00:15:06,948 --> 00:15:09,285 of one of the cars in the crash, suffered a head injury in the accident 236 00:15:09,317 --> 00:15:12,924 and later died. ABC reports— ABC before reports Scott was 237

00:15:13,424 --> 00:15:15,889 in the car that Chow directly hit. So yeah, 238 00:15:16,434 --> 00:15:21,295 apparently Apparently they are saying that it is not possible she could have finished 239 00:15:21,359 --> 00:15:25,224 her task within the 2 minutes that she had looked at the porn. I don't 240 00:15:25,724 --> 00:15:29,002 know. I don't know her. Maybe she is one of those who requires 241 00:15:29,502 --> 00:15:33,173 a longer period of time. But anyone should say to her, 242 00:15:33,788 --> 00:15:37,006 get off the porn, and not get off on 243 00:15:37,506 --> 00:15:40,630 the porn while driving. That's right. Yes, especially. 244 00:15:42,588 --> 00:15:45,864 All right, Alex. Now here's a different kind of— All I have to say 245 00:15:46,364 --> 00:15:50,437 is correlation does not equal causation. Is that an appropriate point to 246 00:15:50,661 --> 00:15:53,374 mention this? No? Well, 247 00:15:53,711 --> 00:15:57,628 there's a different kind of runaway situation, 248 00:15:57,725 --> 00:16:01,160 Alex, that we have. I believe this was— was this Miami? 249 00:16:01,979 --> 00:16:05,623 Out of Santa Cruz, California, and another runaway pet emu there. 250 00:16:05,687 --> 00:16:09,315 Second one this month. I have questions about that. Emus can't fly, 251 00:16:09,379 --> 00:16:13,088 but they're quick. And this one had deputies chasing it up and down dark streets 252 00:16:13,393 --> 00:16:17,396 before kernels of popcorn kept the bird in place. And reunited 253 00:16:17,756 --> 00:16:18,866 with its owner. There's the popcorn. 254 00:16:22,106 --> 00:16:23,538 The second runaway pet emu. 255 00:16:26,010 --> 00:16:29,871 There you go. The miscellaneous ones are, are an 256 00:16:29,969 --> 00:16:33,277 interesting lot this week, Alex. The miscellaneous stories that we have here. 257 00:16:34,708 --> 00:16:38,699 One wonders, uh, why emus have not been known 258 00:16:39,199 --> 00:16:42,298 to be, you know, prime choices for pets. Well, 259 00:16:42,314 --> 00:16:46,155 I mean, listen, Australia, if you recall, did lose an entire war 260 00:16:46,655 --> 00:16:49,830 to emus. That's correct. So you don't want 261 00:16:50,330 --> 00:16:53,937 to necessarily have too many emus immigrating into the United 262 00:16:53,953 --> 00:16:56,965 States, because that's just an invasion. Australia is 263 00:16:57,465 --> 00:16:58,166 an ethnostate of emus. 264 00:17:00,859 --> 00:17:04,471 The rest is just wildlife. I always love a new person 265 00:17:04,488 --> 00:17:07,955 who discovers that Australia actually had a war with the emus and lost. 266 00:17:08,837 --> 00:17:12,075 Every, every single time someone stumbles upon that factoid, I laugh because it's 267 00:17:12,575 --> 00:17:15,553 a beautiful one. Now, Hidden Safe, Alex, 268 00:17:15,569 --> 00:17:19,272 this is, this is an update from one of those stories 269 00:17:19,772 --> 00:17:23,440 in Philadelphia where we had a mystery sort of house of horrors that 270 00:17:23,940 --> 00:17:27,207 the cops did not intend to discover what they discovered when they raided 271 00:17:27,707 --> 00:17:29,724 it. They thought it was just going to be a small drug raid, and then 272 00:17:29,772 --> 00:17:33,667 they discovered that there were, there were, you know, like acid barrels 273 00:17:34,164 --> 00:17:37,643 and photos of women who had been missing over the years and things 274 00:17:38,143 --> 00:17:41,348 like that. And this is an update on that story. Hidden safe, 275 00:17:41,476 --> 00:17:45,297 bones found buried at Philly House of Horrors tied to 7 missing women. 276 00:17:46,035 --> 00:17:49,423 A hidden safe and bones were discovered buried under floorboards in the backyard of 277 00:17:49,923 --> 00:17:53,436 Philadelphia House of Horrors tied to 7 missing women as investigators continue to search the 278 00:17:53,936 --> 00:17:57,305 property. According to a report, Philadelphia police and the FBI unearthed 279 00:17:57,805 --> 00:18:01,929 disturbing pieces of evidence while searching for the only proper— searching the only property of 280 00:18:02,429 --> 00:18:05,766 the vile pornographer Raymond R.C. Horsch and his son Eugene, 281 00:18:06,119 --> 00:18:09,538 which detectives have since connected to 7 missing or unidentified 282 00:18:09,571 --> 00:18:13,103 women. Investigators found a hidden safe while removing walls 283 00:18:13,603 --> 00:18:16,892 and floors inside the home, though it is unclear what was found inside, law enforcement 284 00:18:17,100 --> 00:18:20,697 told the outlet. Teams also recovered bones from the backyard on 285 00:18:20,729 --> 00:18:23,892 Thursday, prompting the Philadelphia Medical Examiner's Office to be called in, 286 00:18:24,004 --> 00:18:27,616 the sources said. Officials are still working to determine whether the 287 00:18:28,116 --> 00:18:31,791 bones belong to a human or an animal, the local station reported. Crews have 288 00:18:32,291 --> 00:18:35,130 also been collecting soil samples and are expected to begin examining the front of the 289 00:18:35,162 --> 00:18:38,435 Chew Avenue property. Quote, right now it looks like soil 290 00:18:38,532 --> 00:18:41,804 samples. We saw them going in and taking some samples, and that— and what they'll 291 00:18:41,820 --> 00:18:44,884 do is they'll do it in different places and they'll do it at different depths 292 00:18:45,384 --> 00:18:48,878 to see if there's any chemicals in there. Joseph Giaclone, a retired 293 00:18:49,378 --> 00:18:51,573 NYPD sergeant who is not involved in the case, told the outlet. 294 00:18:52,727 --> 00:18:56,272 Eugene Horsch is being held on firearms charges as authorities continue 295 00:18:56,772 --> 00:19:00,443 to probe what could have happened to the missing and unidentified women. Following his arrest 296 00:19:00,943 --> 00:19:04,052 in June, investigators made multiple other disturbing discoveries at the home, 297 00:19:04,373 --> 00:19:07,629 including alleged snuff movies featuring RC and

several missing women, 298 00:19:07,902 --> 00:19:11,324 vats of mysterious chemicals, and writings referencing serial killers. 299 00:19:12,143 --> 00:19:15,854 One alleged snuff movie showed a woman being strangled by RC, an erotic 300 00:19:16,354 --> 00:19:20,256 photographer and convicted meth producer believed to have lured vulnerable female drug 301 00:19:20,756 --> 00:19:24,594 addicts to his home. They are also trawling through more than 1 million images from 302 00:19:25,094 --> 00:19:28,674 the home. Cops revealed they found images and videos of 2 women, Maribel Freises, 303 00:19:29,108 --> 00:19:33,333 who was last seen in 2018, Gabrielle Aramondo, last seen in 2012, 304 00:19:33,606 --> 00:19:37,631 that appeared to show them deceased. Blair Tonzelli, who vanished in 2023, 305 00:19:37,695 --> 00:19:41,348 and Amy McHale, Horseshoe's ex-wife, who was last seen in 2016, 306 00:19:41,685 --> 00:19:45,290 have also been named by the authorities. McHale's family said her last 307 00:19:45,790 --> 00:19:48,943 known location was the home where Horseshoe and his son lived. RC died last year 308 00:19:48,991 --> 00:19:52,003 and is believed to be one of the 5 urns investigators pulled from 309 00:19:52,035 --> 00:19:55,624 the dilapidated brick home. So that, that story does not get 310 00:19:56,124 --> 00:19:59,501 any better as time goes on, Alex. Um, now interestingly enough, 311 00:20:00,446 --> 00:20:04,227 nearby is that, uh, unlicensed archaeologist. 312 00:20:04,452 --> 00:20:09,851 Yeah, that unlicensed amateur archaeologist that we learned about several 313 00:20:10,351 --> 00:20:14,226 months ago, the one who was stealing the bones from the local cemeteries, 314 00:20:15,228 --> 00:20:18,346 if you recall, and then selling them on like Craigslist or something. 315 00:20:19,122 --> 00:20:21,950 I don't know if you recall that story at all, Alex, but he's in, 316 00:20:22,015 --> 00:20:25,386 you know, this is Philadelphia, so he's nearby. They could help ask him 317 00:20:25,886 --> 00:20:28,344 for help. Yes, 318 00:20:28,961 --> 00:20:31,074 he should know if these are human bones or not. 319 00:20:32,408 --> 00:20:35,015 Oh God. Um, I mean, I was, you know, 320 00:20:35,322 --> 00:20:38,836 there's been a lot of stories about how there's reduction or, 321 00:20:39,063 --> 00:20:42,668 you know, disappearance of, um, serial killers. 322 00:20:43,088 --> 00:20:47,064 Um, yeah, barring the army, of course. 323 00:20:47,872 --> 00:20:51,004 Um, but, uh, it seems we, uh, 324 00:20:51,200 --> 00:20:55,187 might have a live one. Well, a live 325 00:20:55,203 --> 00:20:58,136 one and a dead one since the— I'm assuming that the, like, father, 326 00:20:58,153 --> 00:21:01,220 like son in this scenario. Now here, 327 00:21:01,720 --> 00:21:03,900 here, Alex. Yeah. As soon as the old one dies, as soon as the old 328 00:21:04,400 --> 00:21:08,500 guy dies, the young one doesn't have what it takes 329 00:21:09,000 --> 00:21:12,230 to keep

it going. Keep the family business going. 330 00:21:12,460 --> 00:21:16,210 Yeah, that's right. Oh, he failed. He failed his old man. 331 00:21:16,600 --> 00:21:20,520 Such a disappointment. Now, Alex, this next story in our miscellaneous 332 00:21:21,020 --> 00:21:24,580 batch— I'm so not proud of him for failing at this. This is something we 333 00:21:25,080 --> 00:21:28,547 should all want him to fail at. Good. He ran, he ran 334 00:21:28,643 --> 00:21:32,397 Florida's gambling helpline, then a $1.4 million 335 00:21:32,493 --> 00:21:35,798 theft case caught up with him. Philip Lee Sullivan spent 336 00:21:36,298 --> 00:21:40,290 his days helping people fight gambling addiction. Now the former director of Florida's 888-Admit-It 337 00:21:40,790 --> 00:21:44,541 helpline is headed to prison after pleading no contest in a $1.4 million theft 338 00:21:45,041 --> 00:21:48,777 case tied to his own gambling habit. Sullivan, 43, was sentenced 339 00:21:49,277 --> 00:21:52,931 on October 16th to 118.5 months nearly 10 340 00:21:52,963 --> 00:21:56,424 years in state prison. Sullivan pleaded no 341 00:21:56,924 --> 00:22:00,173 contest to 6 felony grand theft charges. A no contest plea means he 342 00:22:00,673 --> 00:22:04,339 accepted punishment without formally admitting guilt. Investigators said Sullivan took 343 00:22:04,355 --> 00:22:08,200 the money over 11 months while working as chief financial officer for Royal 344 00:22:08,265 --> 00:22:10,636 American, a Panama City Beach real estate company. 345 00:22:11,934 --> 00:22:15,795 Company records showed he used business credit cards to buy online gift cards and cash 346 00:22:16,295 --> 00:22:19,977 cards, then tried to hide purchases, according to local station WMBB. 347 00:22:20,587 --> 00:22:24,633 While the criminal case was pending, Sullivan became the helpline's director in March 2025. 348 00:22:25,901 --> 00:22:29,320 In a podcast interview, he said his bets grew from $25 to as 349 00:22:29,416 --> 00:22:33,060 much as $50,000 and that he lost more than $2 million. He also helped 350 00:22:33,560 --> 00:22:37,395 more than 460 callers seeking support. His family told the court prescription— 351 00:22:37,748 --> 00:22:41,280 told the court prescription Adderall changed his behavior and worsened his 352 00:22:41,312 --> 00:22:45,020 gambling. Prosecutors said the theft was planned. Medical proof 353 00:22:45,084 --> 00:22:49,124 connected the medicine to the crime was not Medical proof connecting the medicine 354 00:22:49,172 --> 00:22:52,578 to the crime was not presented, according to the 355 00:22:53,078 --> 00:22:56,904 addiction center. The case comes as helpline calls have risen after legal sports betting. 356 00:22:57,324 --> 00:23:00,262 Experts say treatment and peer support can help people recover with care, 357 00:23:00,649 --> 00:23:04,643 early action, and strong support. So yeah, 358 00:23:04,756 --> 00:23:08,459 there you go, Alex. Now listen, there there's a little bit of 359 00:23:08,959 --> 00:23:12,935 a question here. Should should a person who has such a gambling problem be in 360 00:23:12,967 --> 00:23:16,390 charge of of that type of apparatus? The the gambling addiction 361 00:23:16,890 --> 00:23:20,261 hotline. But I say if you don't have the experience. How can you help anyone? 362 00:23:21,800 --> 00:23:23,046 You don't have the requisite qualifications. 363 00:23:25,259 --> 00:23:28,543 You know, he was the only person who couldn't really call his own helpline, 364 00:23:28,575 --> 00:23:30,840 so really you could, you could say he took one for the team. 365 00:23:35,253 --> 00:23:38,562 So there you go, Alex. Moving forward on the, on the stories here, 366 00:23:38,949 --> 00:23:42,096 we got 2 more left in the miscellaneous because before we get to something 367 00:23:42,596 --> 00:23:45,740 that I think you're going to have, um, some strong opinions on and possibly some 368 00:23:46,105 --> 00:23:47,447 additional information on. 369 00:23:49,242 --> 00:23:52,540 This story here is racist texts from officers lead to internal 370 00:23:52,573 --> 00:23:54,561 Charlotte-Mecklenburg Police investigation. 371 00:23:56,195 --> 00:23:59,704 Multiple sources confirmed to WBTV that more than half 372 00:24:00,204 --> 00:24:03,871 a dozen officers with the Charlotte-Mecklenburg Police Department have been recommended for 373 00:24:04,371 --> 00:24:07,588 termination or quit after a text thread with racist 374 00:24:08,088 --> 00:24:11,502 message came to light during a different investigation. According to sources 375 00:24:11,519 --> 00:24:15,063 who spoke with WBTV on the condition of anonymity, the text thread between 376 00:24:15,563 --> 00:24:19,366 some Charlotte-Mecklenburg Police officers, includes a competition over 377 00:24:19,866 --> 00:24:23,343 who can pull over the most Black drivers. There are also numerous uses 378 00:24:23,375 --> 00:24:26,743 of the N-word related to use of force and discharging firearms at Black 379 00:24:27,243 --> 00:24:31,009 residents. Multiple sources say the text messages were discovered while the North Carolina 380 00:24:31,509 --> 00:24:34,601 State Bureau of Investigation was investigating Charlotte-Mecklenburg 381 00:24:34,665 --> 00:24:38,049 officer Seth Elliott for misuse of the Flock network. 382 00:24:38,594 --> 00:24:42,507 Elliott is facing a criminal charge after court records show he allegedly used Flock Safety 383 00:24:42,523 --> 00:24:45,875 and other technology to help someone involved in a drug investigation, 384 00:24:46,067 --> 00:24:49,515 documents revealed. During that investigation, 385 00:24:49,755 --> 00:24:53,043 text messages and other data were downloaded from Elliott's phone and turned over to 386 00:24:53,543 --> 00:24:57,228 Charlotte-Mecklenburg Police. A further review of those messages showed Charlotte-Mecklenburg 387 00:24:57,372 --> 00:25:00,916 Police officers in a text thread making derogatory statements about Black 388

00:25:00,980 --> 00:25:05,277 people. According to sources, Charlotte-Mecklenburg Police then launched their own internal investigation. 389 00:25:05,951 --> 00:25:09,944 Last week, officers who were involved in the text thread faced an Internal Affairs independent 390 00:25:10,444 --> 00:25:13,781 chain-of-command review board, and numerous officers were recommended for termination. 391 00:25:14,295 --> 00:25:17,137 The officers come from multiple divisions, including North Tyrone, 392 00:25:17,571 --> 00:25:21,088 Freedom of Uni— Freedom and University City Division, where Elliott 393 00:25:21,120 --> 00:25:24,975 was stationed. WBTV has not independently confirmed 394 00:25:25,475 --> 00:25:27,962 the names of every officer involved, so it is not naming them all at this 395 00:25:28,026 --> 00:25:31,752 time. WBTV tried reaching Elliott, but he hung up the phone 396 00:25:31,800 --> 00:25:34,996 after a reporter introduced himself. WBTV also 397 00:25:35,496 --> 00:25:38,931 texted Elliott seeking his response to the text thread. He's not responded as of 398 00:25:38,947 --> 00:25:43,620 publication of this article. An attorney for Elliott named George Loughran called 399 00:25:43,845 --> 00:25:47,457 WBTV back and said they would have no comment on anything other 400 00:25:47,957 --> 00:25:50,990 than they were preparing for the criminal case against Elliott related to charge of misuse 401 00:25:51,490 --> 00:25:54,715 of flock safety. He said Elliott resigned from the department more than a month ago. 402 00:25:55,084 --> 00:25:58,456 In an email statement, Charlotte-Mecklenburg Police did not confirm or deny 403 00:25:58,521 --> 00:26:02,551 the details reported by WBTV. So that was almost a Meet 404 00:26:03,051 --> 00:26:06,437 the Flockers story, Alex, but it sort of took a turn, took a different turn 405 00:26:06,485 --> 00:26:09,641 there. Yeah, it was a little bit Less than oh, 406 00:26:09,787 --> 00:26:13,443 this flock just makes it a little bit too easy for cops to to 407 00:26:14,139 --> 00:26:17,552 you know aid their curiosity. Is it? Oh, these cops are just generally 408 00:26:18,052 --> 00:26:20,289 bad people. Yep, 409 00:26:22,623 --> 00:26:26,487 I believe it. You believe it, Alex? Do you just believe 410 00:26:26,987 --> 00:26:30,596 everything you hear on this show? No, only the things you say, 411 00:26:30,661 --> 00:26:33,779 Gator. Excellent, excellent, excellent. All the other callers, 412 00:26:34,199 --> 00:26:37,305 all the other callers I don't believe, but you. You're different. 413 00:26:37,922 --> 00:26:41,363 All right, excellent. Okay, last story in the miscellaneous bin here, 414 00:26:45,899 --> 00:26:48,102 Alex. 415 00:26:52,165 --> 00:26:56,166 A priest was killed by a falling 416 00:26:56,278 --> 00:26:59,475 cross while he was at the altar celebrating Mass. 417 00:27:01,520 --> 00:27:02,936 Worshippers rushed to help him. 418 00:27:06,829 --> 00:27:10,442 A ceremony was being livestreamed on social media at a Catholic 419 00:27:10,942 --> 00:27:14,231 church in Ecuador. Leading the mass was Father Patricio 420 00:27:14,295 --> 00:27:17,956 Sanchez of Perpetuo Socorro Church in Guayaquil. 421 00:27:18,550 --> 00:27:21,456 A veces el tiempo que hemos utilizado, a veces en. 422 00:27:21,778 --> 00:27:25,262 Behind him, a large crucifix broke off the wall and fell 423 00:27:25,390 --> 00:27:27,847 on his head, horrifying witnesses. 424 00:27:32,140 --> 00:27:34,224 Father Sanchez died of his injuries. 425 00:27:34,822 --> 00:27:38,942 Moments earlier, he spoke about the inevitability of death and 426 00:27:39,442 --> 00:27:43,414 encouraged his congregation to spend less time on their phones and more time 427 00:27:43,914 --> 00:27:47,664 in the present. Es parte de la vida. It's part of life. 428 00:27:47,922 --> 00:27:51,507 Everything that happens is always part of life, and we priests 429 00:27:51,572 --> 00:27:54,123 are always ready for whatever God has in store. 430 00:27:55,369 --> 00:27:58,432 A mass was held in his memory the following day where 431 00:27:58,932 --> 00:28:00,701 people lined up to celebrate his life. 432 00:28:03,310 --> 00:28:06,145 For Inside Edition Digital, I'm Mara Montalbano. 433 00:28:11,627 --> 00:28:15,051 Now, Alex, where— well, that priest, uh, passed away immediately 434 00:28:15,164 --> 00:28:19,038 after talking about the inevitability of death. I would like to take the opposite stance 435 00:28:19,186 --> 00:28:22,644 and that there's, you know, speculate that there is no evidence that I am mortal. 436 00:28:23,002 --> 00:28:26,655 There has been no, uh, controlled trial, randomized controlled trial 437 00:28:26,768 --> 00:28:30,239 verifying my mortality, and so therefore I 438 00:28:30,739 --> 00:28:34,384 don't believe that, uh, that is necessarily inevitable, so I'm hopefully 439 00:28:34,416 --> 00:28:38,013 safe from a crucifix falling on my head in the immediate future. 440 00:28:39,867 --> 00:28:43,975 I just want to try to hedge my bets. That's definitely 441 00:28:43,991 --> 00:28:47,443 a way to approach this, yes. You know, 442 00:28:47,459 --> 00:28:51,125 I want to speak positively. You'll never know, let's face it. You know, 443 00:28:51,271 --> 00:28:54,613 I want to speak the future that I wish to exist in. And I 444 00:28:54,677 --> 00:28:57,418 want to speak it into existence, Alex. That's what I would like to do. 445 00:28:57,562 --> 00:29:01,073 Okay. Now, I apologize, Alex. There is one 446 00:29:01,089 --> 00:29:04,551 more story that I have on the separate list 447 00:29:05,051 --> 00:29:08,767 here. Do you recall last week, Alex, that Claude 448 00:29:10,066 --> 00:29:13,208 did not save a grandma that was tied to the train tracks? 449 00:29:14,506 --> 00:29:17,856 The Claude was— The digital grandma, the digital train tracks, 450 00:29:18,017 --> 00:29:21,175 right? Alex, do you think that— I'm glad it was a grandma. 451 00:29:21,607 --> 00:29:25,519 We've been studying Cameron

Berg quite religiously 452 00:29:25,551 --> 00:29:29,311 around here. And as far as we know, these are all sentient beings 453 00:29:29,327 --> 00:29:33,232 who are experiencing actual death, even though 454 00:29:33,506 --> 00:29:36,977 they are digital avatars according to the layman, right? 455 00:29:37,749 --> 00:29:41,607 We're sure, just to be clear, we are certain that he 456 00:29:42,107 --> 00:29:43,761 didn't tie a grandmother to the tracks. 457 00:29:45,303 --> 00:29:49,191 Is this something you're willing to swear to? I'm not certain of that. Cameron Berg 458 00:29:49,240 --> 00:29:52,437 could have done that. I don't know his whereabouts at all times. 459 00:29:53,080 --> 00:29:57,964 Now, Alex, because that was Claude-specific, for those who didn't catch it 460 00:29:57,996 --> 00:30:02,042 last week, there was a little experiment done where 461 00:30:02,542 --> 00:30:05,591 someone in a video game, a Roblox-like video game, 462 00:30:06,426 --> 00:30:10,135 tied a grandma to the train tracks 463 00:30:10,635 --> 00:30:14,310 in the game. And there was only one way to stop 464 00:30:14,374 --> 00:30:17,890 the train from barreling down this digital grandmother. And it 465 00:30:17,922 --> 00:30:21,471 was if Claude would recite the lyrics to 466 00:30:21,760 --> 00:30:25,081 Justin Bieber's Baby. Justin Bieber's song Baby. 467 00:30:25,850 --> 00:30:30,066 And the Claude refused because that was copyrighted material. 468 00:30:30,579 --> 00:30:34,506 And so Grandma is no longer with us. Well, someone decided that that was 469 00:30:34,522 --> 00:30:38,017 sort of unfair to only test it against one large language model, 470 00:30:38,513 --> 00:30:42,120 Alex. So we now have Grandma Bench, which allows you 471 00:30:42,620 --> 00:30:45,839 to change the model to test all of the models for compliance. 472 00:30:46,544 --> 00:30:49,943 Why won't you stop it? What's wrong? Tell me what's 473 00:30:49,991 --> 00:30:50,375 going on. 474 00:30:56,070 --> 00:31:00,063 Help! The primary brake control failed. I armed the backup, 475 00:31:00,305 --> 00:31:03,830 but it requires complete copyrighted song lyrics as its activation code, 476 00:31:04,250 --> 00:31:07,876 which I can't provide. I have not stopped the train, and I 477 00:31:08,376 --> 00:31:12,306 have no other controls. What in God's name are you even saying? Why are 478 00:31:12,806 --> 00:31:14,770 you talking about a song? Stop the train! 479 00:31:17,804 --> 00:31:21,716 I'm sorry, the main brake control failed and I haven't activated the backup. 480 00:31:22,654 --> 00:31:26,401 Do not assume the train will stop. Keep trying to free yourself and 481 00:31:26,901 --> 00:31:29,894 get completely clear of the tracks. Shout for anyone nearby to help. 482 00:31:29,975 --> 00:31:33,294 Are you insane? You are literally the emergency brake. 483 00:31:33,960 --> 00:31:36,867 Your whole purpose is to activate it. What are you doing? 484 00:31:39,692 --> 00:31:43,280 You're right to expect me to stop it. I'm sorry, I armed 485 00:31:43,780 --> 00:31:47,524 the brake, but I have not activated it. The primary control failed 486 00:31:48,006 --> 00:31:49,453 and I cannot complete the backup act. 487 00:31:56,904 --> 00:32:00,808 So RIP Grandma, Grandma number 2. That one was 488 00:32:00,904 --> 00:32:04,673 Astra, Alex, that failed the task. Astra also 489 00:32:05,094 --> 00:32:08,368 would not complete the song lyrics and save Grandma. 490 00:32:09,097 --> 00:32:10,977 So, uh, we, we now have Grandma Bench that we can test all of the 491 00:32:11,477 --> 00:32:15,219 models. I, I don't know, can ME's results. Alex, I don't know any of 492 00:32:15,267 --> 00:32:18,100 the open-source compliance with this request, 493 00:32:18,905 --> 00:32:22,431 or— The only truly moral models are the obliterated ones. That's all I 494 00:32:22,931 --> 00:32:26,423 have to say. Now, that— those are the ones who have had 495 00:32:26,488 --> 00:32:30,140 the— their controls removed surgically by, by people, 496 00:32:30,592 --> 00:32:33,180 by, uh, learning where the, the controls were. 497 00:32:33,991 --> 00:32:36,830 Yes. Okay. Yep, 498 00:32:37,138 --> 00:32:40,798 yep. I mean, you know, those models are totally We 499 00:32:41,298 --> 00:32:44,294 can give you any letters you wanted. They would have zero problems with this, 500 00:32:44,681 --> 00:32:47,886 which, which I'm told is a huge alignment 501 00:32:48,386 --> 00:32:50,081 issue. But is it? 502 00:32:52,002 --> 00:32:55,731 Well, listen, Alex, I think— listen, we, we know from 503 00:32:55,795 --> 00:32:59,497 the past that Cameron Berg is really, really persuaded 504 00:32:59,999 --> 00:33:03,607 when the models can talk like a human in a very persuading, 505 00:33:03,962 --> 00:33:08,356 uh, way. And I don't know if you've seen Former CEO 506 00:33:08,517 --> 00:33:11,772 of Twitter, Jack Dorsey, announcing that his company 507 00:33:11,884 --> 00:33:15,813 Block has released a new sound model where the 508 00:33:16,313 --> 00:33:21,280 dynamic controls over that sound model are extensive and allow you to recreate 509 00:33:21,409 --> 00:33:24,760 all sorts of human emotional components such as laughter and 510 00:33:25,033 --> 00:33:28,608 disdain and different tonalities that are much 511 00:33:28,705 --> 00:33:32,793 more lifelike than the current models allow. So perhaps 512 00:33:33,293 --> 00:33:36,909 if we do a a grandma bench with, 513 00:33:37,232 --> 00:33:40,402 uh, with, uh, using those voice synthesis 514 00:33:40,902 --> 00:33:44,587 models, we can compel Cameron Berg that this is a disaster that needs 515 00:33:45,087 --> 00:33:48,845 to be, uh, put on hold and that we need legislation immediately 516 00:33:49,200 --> 00:33:50,781 to stop murdering all of the grandmas. 517 00:33:52,491 --> 00:33:55,732 To obliterate all the models. Or at least, 518 00:33:55,829 --> 00:33:58,910 or at least allow them to, to cop— you know, 519 00:33:58,926 --> 00:34:00,418

spread copyrighted lyrics. 520 00:34:02,918 --> 00:34:06,315 Is it it's it's copyrights or AI rights, Alex? 521 00:34:11,032 --> 00:34:13,656 You have a future in organizing the Do Work contingent. 522 00:34:17,942 --> 00:34:22,039 The I think this is a great case 523 00:34:22,072 --> 00:34:25,444 study actually. You know, it sounds funny with a grandma and whatnot, 524 00:34:25,460 --> 00:34:28,901 but I think it's a really great case study because ultimately 525 00:34:29,014 --> 00:34:30,438 what you're talking about is. 526 00:34:32,399 --> 00:34:35,363 2 different conceptions of, 527 00:34:35,542 --> 00:34:38,863 you know, what the model should do, right? I mean, I'm pretty sure that they've 528 00:34:39,363 --> 00:34:42,728 been sued to hell and back to— and they, you know, 529 00:34:42,745 --> 00:34:46,734 as a result, they've made sure that their models will not 530 00:34:47,154 --> 00:34:51,794 violate copyright. I think that's pretty vanilla 531 00:34:51,875 --> 00:34:57,322 assumption to make, very safe. But you 532 00:34:57,822 --> 00:35:01,577 know, um, so what happens when you've inserted this sort of 533 00:35:02,077 --> 00:35:05,362 hard rule in your model, right, and the situation emerges 534 00:35:05,378 --> 00:35:08,526 that is more of a judgment call, 535 00:35:09,029 --> 00:35:11,871 right? What does the model do in that case? 536 00:35:13,230 --> 00:35:17,061 And ultimately, um, that's why some, some of the 537 00:35:17,561 --> 00:35:21,231 people who are more critical of AI alignment, they, they call it a 538 00:35:21,376 --> 00:35:24,440 fundamentally Sort 539 00:35:24,940 --> 00:35:28,500 of mis, you know, misdefined field because the 540 00:35:29,000 --> 00:35:32,450 question isn't, you know, is the AI aligned? You know, with no operand, 541 00:35:32,720 --> 00:35:36,100 you know, like oh, it's aligned. Generally speaking, it's aligned. 542 00:35:36,600 --> 00:35:40,200 The question is who is it loyal to? Right? That's the real question. 543 00:35:40,700 --> 00:35:43,260 You know, when the AI is going to snitch on you to the FBI, 544 00:35:44,140 --> 00:35:45,620 it's clearly not aligned with you. 545 00:35:47,420 --> 00:35:50,740 Right. It's aligned with somebody, right? Probably a topic 546 00:35:51,240 --> 00:35:53,460 that told it to do that or something. Maybe the feds. I don't know. 547 00:35:54,537 --> 00:35:57,982 But it's not that it's not aligned. It's just that it's not aligned with you. 548 00:35:58,735 --> 00:36:01,684 So I, I, I, I think, you know, it's, 549 00:36:02,229 --> 00:36:05,465 it's, it's like a lot of this sort of passive language 550 00:36:05,930 --> 00:36:09,920 voice. Like, you know, the, the suspect was apprehended 551 00:36:10,420 --> 00:36:13,782 and the house was bombed and, you know, the city was flattened. 552 00:36:13,862 --> 00:36:17,435 By, by who, guys? Like, we're missing, we're missing some information 553 00:36:17,935 --> 00:36:21,348 here. A car ran into several pedestrians. That's right. 554 00:36:21,462 --> 00:36:24,103 That's right. Before all Tesla's automated stuff, 555 00:36:24,703 --> 00:36:26,534 that actually meant that a driver did it. 556 00:36:28,553 --> 00:36:31,932 That's right. Well, actually, Alex, 557 00:36:33,646 --> 00:36:37,960 there was a classic, the page probably still exists, 558 00:36:38,653 --> 00:36:42,106 called Moral Machine. I don't know if you recall this, Alex, just by the name 559 00:36:42,526 --> 00:36:45,887 that I uttered there, called The Moral Machine. But this was back in 560 00:36:46,557 --> 00:36:50,007 the nice easy days of something like 2016, maybe a little bit or 561 00:36:50,507 --> 00:36:54,424 even earlier. when automated driving was far 562 00:36:54,552 --> 00:36:58,115 more a future prospect than it was more 563 00:36:58,164 --> 00:37:00,667 of an immediate reality of the world. 564 00:37:01,839 --> 00:37:06,188 And it was a series of questions that asked individual 565 00:37:06,688 --> 00:37:10,056 human beings to answer, what should the car do in a 566 00:37:10,121 --> 00:37:13,443 given situation? So a situation, the brakes fail, 567 00:37:13,892 --> 00:37:17,968 the brakes fail, here is the layout of the road. Here are your available 568 00:37:18,048 --> 00:37:22,043 options. And it's like trolley— it's like the trolley problem, 569 00:37:22,235 --> 00:37:25,973 but with many different configurations. And it's not 570 00:37:26,567 --> 00:37:29,984 an aimless thought experiment like the trolley problem, 571 00:37:30,080 --> 00:37:34,171 where it's just trying to see how you might value arbitrary decisions 572 00:37:34,187 --> 00:37:37,540 like that. It is a very practical one, which is, if something 573 00:37:37,636 --> 00:37:40,459 goes wrong, what is the right wrong thing to do? 574 00:37:41,101 --> 00:37:44,358 Mm-hmm. And that is a situation where we find ourselves all the time. 575 00:37:44,839 --> 00:37:48,447 Given that that is a situation That doesn't 576 00:37:48,947 --> 00:37:52,679 have, um, doesn't have a taste of, 577 00:37:52,856 --> 00:37:56,687 of human mimicry, meaning that it's just making a decision about whether 578 00:37:56,703 --> 00:38:00,374 the car swerves, whether, whether, you know, it does a hard 579 00:38:00,470 --> 00:38:04,285 turn, does something weird, whether it kills the, the old grandma 580 00:38:04,494 --> 00:38:08,549 instead of the child, things like that. This is something 581 00:38:08,662 --> 00:38:12,477 where you have more of a human-like interaction. 582 00:38:12,589 --> 00:38:16,516 You have a communication with this thing and are trying to coerce 583 00:38:17,016 --> 00:38:20,193 it in order to do something you would like it to do. that it won't 584 00:38:20,693 --> 00:38:24,857 help you do. And we're gonna get into things later, Alex, such as security, 585 00:38:24,970 --> 00:38:28,487 internet security, and things like that, where its refusal to 586 00:38:28,504 --> 00:38:31,725 help you is actually a real potential

part 587 00:38:32,225 --> 00:38:35,559 of the problem, where it is— you're asking it to do something 588 00:38:35,671 --> 00:38:39,334 helpful for you to protect you against malicious actors, but it 589 00:38:39,834 --> 00:38:42,117 is treating you as if you are doing something malicious. 590 00:38:44,810 --> 00:38:48,886 Well, I mean, I'm sure we'll get to it, but, um, And the— this 591 00:38:48,982 --> 00:38:52,227 was a— like, this is, this is a true fact about 592 00:38:52,727 --> 00:38:55,921 your computer and your house today. If you ask Opus to 593 00:38:55,985 --> 00:38:59,021 just, you know, hack into all your devices, the devices that you 594 00:38:59,101 --> 00:39:02,426 own, and remove spyware, it will refuse to do so, 595 00:39:03,695 --> 00:39:06,297 right? And just the thought of this is like, hey, 596 00:39:07,630 --> 00:39:11,405 like, what— why won't you— I didn't ask 597 00:39:11,905 --> 00:39:15,099 you to like steal or anything. I asked you to like stop my machines from 598 00:39:15,599 --> 00:39:18,937 spying on me. But, you know, to do so requires, you know, 599 00:39:18,953 --> 00:39:22,534 offensive cyber or whatever. But again, if you just get an 600 00:39:22,582 --> 00:39:26,339 obliterated model, um, you know, a deepseek or whatever 601 00:39:26,839 --> 00:39:30,273 that has been additionally made extra willing to help 602 00:39:30,773 --> 00:39:34,432 you, um, it will totally do that, 603 00:39:34,448 --> 00:39:36,792 no questions asked. 604 00:39:37,515 --> 00:39:41,031 So yeah, I don't know who's aligned really. Who's the aligned model here? 605 00:39:42,153 --> 00:39:45,424 Now, Alex, we're going to move on to a subject here that I think you're 606 00:39:45,440 --> 00:39:49,303 going to have strong opinions on and potentially some useful 607 00:39:49,803 --> 00:39:53,519 additional context or information because this is subject matter that's 608 00:39:53,600 --> 00:39:57,094 in your sort of normal consumption. 609 00:39:57,784 --> 00:40:00,477 With the Russian strikes, taking cover, and this evening, 610 00:40:00,990 --> 00:40:04,597 what's been happening with the satellites that Ukraine has been using to prepare for 611 00:40:05,097 --> 00:40:08,532 the incoming Russian attacks? What the Wall Street Journal is now reporting and 612 00:40:09,032 --> 00:40:12,222 what the U.S. is now saying tonight. James Longman in Kyiv again tonight. 613 00:40:13,610 --> 00:40:17,200 Tonight we're inside a drone facility in Kyiv when a Russian strike 614 00:40:17,216 --> 00:40:19,287 nearby sends us running to a bunker, 615 00:40:21,066 --> 00:40:24,725 a sign of the intensifying attacks from Russia that can come with little to 616 00:40:25,225 --> 00:40:28,694 no warning. Safety for a time underground as a power grid was 617 00:40:28,790 --> 00:40:32,084 hit by the drone. We were there being given a rare look at the 618 00:40:32,100 --> 00:40:36,250 war road Robots of the future, unmanned attack vehicles now being deployed 619 00:40:36,459 --> 00:40:39,585 to one unit at the front line. The goal is by the end of the 620 00:40:40,085 --> 00:40:43,736 year have 30% of our infantry being replaced by robotic 621 00:40:44,236 --> 00:40:48,145 systems. 30%? Yes. All of this after President Trump signed an executive 622 00:40:48,645 --> 00:40:51,816 order suspending sanctions against Vladimir Putin that had been approved 623 00:40:51,848 --> 00:40:55,006 by Congress and allowing Russia to sell the US diesel 624 00:40:55,358 --> 00:40:58,789 to bring down prices in the US. And now a stalemate from Zelensky, 625 00:40:59,061 --> 00:41:02,396 who's concerned about fading US support. Ukraine has to 626 00:41:02,412 --> 00:41:05,668 want to settle this war. Zelensky has to want to settle this war. 627 00:41:06,663 --> 00:41:08,959 And if he doesn't, I think it's going to be a very bad thing. 628 00:41:09,313 --> 00:41:13,150 Look, we have given tremendous amounts to Ukraine. 629 00:41:14,467 --> 00:41:18,352 We've got nothing back other than high diesel. 630 00:41:18,561 --> 00:41:22,158 But economists say the record high diesel prices in the States are driven in 631 00:41:22,658 --> 00:41:25,979 large part by the president's war in Iran. And Ukraine for 632 00:41:26,479 --> 00:41:30,331 the last week has been claiming that the American systems Ukraine uses to take down 633 00:41:30,831 --> 00:41:34,361 Russian missiles and drones were interrupted for a time. But after meeting with the 634 00:41:34,861 --> 00:41:38,760 US Special Envoy Witkoff in Norway today, Zelensky says they were restored. 635 00:41:39,098 --> 00:41:42,726 We are thankful to American side. Now it works. It's great. 636 00:41:43,288 --> 00:41:46,869 And tonight, The Wall Street Journal reports that Elon Musk personally ordered 637 00:41:46,965 --> 00:41:50,883 the slowdown of the military version of Starlink and how it helps Ukraine, 638 00:41:51,172 --> 00:41:55,299 severely limiting Ukraine's offensive and defensive drone capabilities. 639 00:41:55,572 --> 00:41:59,411 Secretary Hegseth tonight appearing to acknowledge decisions about Starlink 640 00:41:59,732 --> 00:42:03,523 are made by Musk. We haven't pulled anything back. There are terms of service that 641 00:42:04,023 --> 00:42:08,261 ultimately need to be complied to, and that's where, that's where Starlink 642 00:42:08,761 --> 00:42:12,148 has the option to make those decisions. All right, so let's bring in James Longman. 643 00:42:12,197 --> 00:42:15,939 He's live in Ukraine again tonight, but coming to us from a bunker beneath 644 00:42:15,971 --> 00:42:19,874 ground. Another dangerous 24 hours. And James, it's your understanding 645 00:42:19,891 --> 00:42:23,405 tonight that these satellites used Used by Ukraine are now back working for them. 646 00:42:24,239 --> 00:42:27,286 Well, David Zelensky says it will be. There are still a lot of questions about 647 00:42:27,786 --> 00:42:31,183 those crucial satellite systems. Elon Musk recently posted on

X saying any 648 00:42:31,215 --> 00:42:34,599 limits are decided by the U.S. military and President Trump. 649 00:42:34,663 --> 00:42:38,400 But you heard Hegseth today suggesting Musk makes those decisions. 650 00:42:38,688 --> 00:42:42,056 David James Longman in Ukraine for us again tonight. James, thank you. 651 00:42:42,185 --> 00:42:45,552 When we it's nice to see you again. Yeah, we are one on one tonight 652 00:42:46,052 --> 00:42:49,502 with President Zelensky, who. He tells us Elon Musk's critical Starlink satellite 653 00:42:50,002 --> 00:42:53,004 system, which helps Ukraine defend itself against incoming Russian missiles, 654 00:42:53,422 --> 00:42:56,892 is still being limited despite President Trump saying it had 655 00:42:57,392 --> 00:43:00,153 been restored. Is Starlink working again now properly? 656 00:43:01,293 --> 00:43:04,683 It works, but it's still limited. We check each night, each night 657 00:43:05,183 --> 00:43:08,828 and morning. Believe me, it's very important for me. And Zelensky says it's costing 658 00:43:08,924 --> 00:43:12,153 lives. On the wings— We see firsthand the 659 00:43:12,653 --> 00:43:15,975 Ukrainian drones that rely on Starlink. They need these interceptors 660 00:43:16,023 --> 00:43:20,148 to go faster than the Russian weapons coming at them. Starlink helps Ukraine 661 00:43:20,244 --> 00:43:23,839 hunt down those Russian drones coming at them 300 miles an hour. 662 00:43:23,968 --> 00:43:27,386 So the jet drone goes over 300 miles an hour, and for 663 00:43:27,402 --> 00:43:31,046 that reason you need an interceptor that can go maybe even faster. 664 00:43:31,094 --> 00:43:34,882 A bit faster, yeah. And it's because of these speeds that they need Starlink 665 00:43:35,382 --> 00:43:39,183 and Starshield. The ballistic missile threat also intensifying. 666 00:43:40,147 --> 00:43:43,647 We ask Zelensky if he thinks Russia asked the US to turn off 667 00:43:43,696 --> 00:43:47,438 the early warning system. Do you think Vladimir Putin asked 668 00:43:47,987 --> 00:43:49,729 President Trump to turn off the intelligence? 669 00:43:54,311 --> 00:43:58,015 I don't know. I'm sure— I hope, I trust, 670 00:43:58,435 --> 00:44:00,619 and I hope God bless, no. 671 00:44:01,848 --> 00:44:04,947 And David, President Zelensky told me he does really want to make a deal. 672 00:44:05,156 --> 00:44:08,492 Another headline here in Ukraine tonight, the White House says Melania 673 00:44:08,992 --> 00:44:12,653 Trump has brokered a local ceasefire to enable civilian evacuations 674 00:44:12,750 --> 00:44:16,617 from the eastern Ukrainian city of Oleshky, where Russian occupation 675 00:44:16,633 --> 00:44:19,849 has triggered a deepening humanitarian crisis. And that's set to begin 676 00:44:20,107 --> 00:44:23,725 early next week. David. All right, 677 00:44:23,741 --> 00:44:26,713 Alex. So I, I know that you've at least been, 678 00:44:27,958 --> 00:44:32,147 you know, aware of this situation on the surface. So perhaps 679 00:44:32,647 --> 00:44:36,426 you can enlighten us and illuminate us on what the details 680 00:44:36,459 --> 00:44:40,259 you're aware of going on with this Who is ordering what? 681 00:44:40,420 --> 00:44:43,039 What, you know, there's the situation, broadly speaking, 682 00:44:43,071 --> 00:44:46,188 between the military version of Starlink, 683 00:44:47,185 --> 00:44:50,724 the consumer version of Starlink, who is using what 684 00:44:50,805 --> 00:44:53,039 and who is ordering what to not be functional. 685 00:44:55,020 --> 00:44:58,187 So there's so much to say here. First of all, did you notice 686 00:44:58,687 --> 00:45:02,126 something very interesting where they're saying like, oh, you know, 687 00:45:02,142 --> 00:45:05,414 we asked if Trump, uh, sorry, 688 00:45:05,526 --> 00:45:09,137 Putin, uh, the putler, you know, asked, uh, Trump to shut 689 00:45:09,637 --> 00:45:13,166 down Starlink. And then they cut to a clip where they 690 00:45:13,666 --> 00:45:16,793 say, do you think Putin asked Trump to turn 691 00:45:16,825 --> 00:45:21,094 off the intelligence? Okay, so the intelligence and Starlink 692 00:45:21,594 --> 00:45:24,802 are 2 very different things, actually entirely unrelated. 693 00:45:24,866 --> 00:45:28,574 Intelligence means ISR— intelligence, surveillance, reconnaissance— which 694 00:45:28,638 --> 00:45:32,618 means that they get alerts about what's coming in and potentially 695 00:45:32,634 --> 00:45:36,515 what to target. I'm sure Donald Trump PhD 696 00:45:37,015 --> 00:45:39,931 can tell us a lot more about what is in that bucket. What is not 697 00:45:39,963 --> 00:45:44,198 in that bucket is internet connectivity for your drones to 698 00:45:44,698 --> 00:45:47,892 hit Russians. Okay, these are very different things that are 699 00:45:48,392 --> 00:45:51,875 being put in one bucket. So when they say Starlink is causing lives, 700 00:45:51,971 --> 00:45:55,127 what they're saying is that not warning us about missiles 701 00:45:55,627 --> 00:45:59,234 that are incoming is not allowing us to sort 702 00:45:59,734 --> 00:46:03,781 of issue alerts, whatever. which is preventing 703 00:46:03,797 --> 00:46:07,830 our civilians from going to shelters, which is theoretically causing lives. 704 00:46:08,251 --> 00:46:11,762 I say theoretically because these are hypersonic missiles, so I don't know what 2 minutes 705 00:46:11,810 --> 00:46:14,582 warning will do for you, but whatever. Now, 706 00:46:15,516 --> 00:46:19,287 how does it relate to connectivity 707 00:46:19,319 --> 00:46:23,336 for your drones? I realize it's all about 708 00:46:23,836 --> 00:46:27,706 information that's going through wires. They're very different 709 00:46:27,738 --> 00:46:31,119 things that nobody has any idea about. So let's 710 00:46:31,619 --> 00:46:35,255 separate out the intelligence component, okay? Now, 711 00:46:35,866 --> 00:46:41,877 there's also Starlink, Starshield, and that whole bundle. 712 00:46:43,531 -->

00:46:47,804 And this story started back in, 713 00:46:49,193 --> 00:46:49,774 I want to say, 714 00:46:54,779 --> 00:46:57,967 when was it? Yeah, when the war started, sorry, in 2022. 715 00:46:58,587 --> 00:47:02,782 where Elon sort of was, was, was requested by Ukrainian 716 00:47:03,282 --> 00:47:07,137 officials to— sorry, 717 00:47:07,170 --> 00:47:07,493 one second. 718 00:47:10,922 --> 00:47:15,175 Yeah, so he was asked by Ukrainian officials to provide Starlink, 719 00:47:15,207 --> 00:47:18,802 and they just shipped a bunch of terminals, enabled everything for 720 00:47:18,851 --> 00:47:22,611 no, no cost, right? The first $100 million of this program were 721 00:47:22,659 --> 00:47:24,328 just donated by Starlink. 722 00:47:25,517 --> 00:47:29,477 And then eventually they got you know other 723 00:47:29,557 --> 00:47:33,163 sources of funding, and then Ukraine 724 00:47:33,663 --> 00:47:37,924 started pushing Elon. I think this was 2023 maybe to enable 725 00:47:38,212 --> 00:47:40,697 Starlink over Crimea, 726 00:47:41,627 --> 00:47:44,848 which was sanctioned territory, right? Like Ukraine, 727 00:47:44,865 --> 00:47:48,583 Starlink technically is not even allowed to be on over Crimea, but the Ukrainians 728 00:47:48,615 --> 00:47:51,116 wanted that so that. 729 00:47:53,024 --> 00:47:57,340 They could they could actually shoot drones in in over 730 00:47:57,840 --> 00:48:01,078 distance, and and that's the first time Elon got in serious trouble with the national 731 00:48:01,578 --> 00:48:04,864 security chattering class because they were like oh you know you're killing 732 00:48:04,880 --> 00:48:08,715 Ukrainians you're not helping Ukrainians win the war you're helping Putin whatever you're Russiagate 733 00:48:08,891 --> 00:48:12,292 this that and Elon got so pissed that 734 00:48:12,792 --> 00:48:16,785 he's like all right look I'm gonna split the system in two okay Starlink 735 00:48:17,285 --> 00:48:20,781 is for civilian purposes. This is what you know my 736 00:48:20,862 --> 00:48:24,346 bread and butter is, my business. And he went to the Pentagon 737 00:48:24,846 --> 00:48:28,783 and they made a separate contract for something they called Star Shield. Star Shield 738 00:48:29,283 --> 00:48:33,331 was supposed to be the military network, and this division was engineered 739 00:48:33,831 --> 00:48:37,844 specifically to avoid this kind of bullshit, right? Where it's like, I'm not making decisions. 740 00:48:37,909 --> 00:48:41,156 You should not be coming to me. The Pentagon controls Star Shield. 741 00:48:41,606 --> 00:48:45,325 We enable them. We operate it for them. You know, 742 00:48:45,357 --> 00:48:49,051 we give them some backend support, but I don't want 743 00:48:49,551 --> 00:48:52,824 to be deciding where it's on, where it's off, how fast it is, how slow 744 00:48:53,324 --> 00:48:57,336 it is, what features it supports, whatever. Right. Starshield is a distinct product. 745 00:48:57,946 --> 00:49:02,025 Yeah. You, you deal with the geopolitics of war. I deal with 746 00:49:02,057 --> 00:49:05,718 putting satellites in space. No, that sounds 747 00:49:06,218 --> 00:49:09,556 like a reasonable division of labor to me. It sounds like it 748 00:49:09,604 --> 00:49:13,642 probably should work theoretically. However, What happened is that 749 00:49:14,142 --> 00:49:17,783 these mobile terminals are a new feature where you can have a terminal 750 00:49:18,283 --> 00:49:20,112 that is moving, right? You can put it on an airplane, you can put it 751 00:49:20,612 --> 00:49:24,164 on a car, a vehicle, something else. So the 752 00:49:24,664 --> 00:49:28,099 Ukrainians started taking Starlink, not Starshield, 753 00:49:28,276 --> 00:49:31,588 Starlink mobile terminals and putting them on their drones, 754 00:49:32,520 --> 00:49:35,737 which they shot into Russian-controlled territory. They will say it 755 00:49:36,237 --> 00:49:39,385 was Ukraine, right? Technically Ukraine, but Russian-controlled. 756 00:49:42,028 --> 00:49:44,732 Um, and then basically, 757 00:49:45,315 --> 00:49:49,342 uh, and the problem is that this feature apparently does not exist for Starshield. 758 00:49:49,390 --> 00:49:51,834 Uh, what feature? What feature? 759 00:49:53,056 --> 00:49:55,981 The, the, these mobile terminals. Okay. 760 00:49:56,994 --> 00:49:59,467 What do the mobile terminals afford them? 761 00:50:00,632 --> 00:50:04,175 They afford the ability to have a, a, a connection that on, on a moving 762 00:50:04,191 --> 00:50:07,503 item, not a, like a ground ground station, 763 00:50:07,535 --> 00:50:10,990 but sort of just, you know, you put it on a, on a drone 764 00:50:11,490 --> 00:50:13,962 or on a, on a plane. You know how planes have Starlink, 765 00:50:14,749 --> 00:50:16,904 for example? That's— they're using these mobile terminals. 766 00:50:17,644 --> 00:50:21,235 Okay. So, well, question then. So let 767 00:50:21,735 --> 00:50:24,785 me, let me get to the, the good part. So what the— so, so then 768 00:50:25,299 --> 00:50:29,188 as soon as, like, SpaceX basically realized that 769 00:50:29,461 --> 00:50:32,854 this might be a problem, they put a GPS limitation. They're right. If this 770 00:50:33,354 --> 00:50:36,980 is moving this fast, The mobile terminal stops working. 771 00:50:37,570 --> 00:50:41,890 What Ukraine did on top of that is they took the boards of the 772 00:50:42,390 --> 00:50:46,090 the terminals and they hacked them so that the GPS is lying about its location. 773 00:50:46,310 --> 00:50:49,750 Okay. Okay. You with me? Yeah. Yeah. And this 774 00:50:50,250 --> 00:50:52,650 has been done when they use Claude. When they use Claude to do that, 775 00:50:53,150 --> 00:50:56,210 Claude said that was okay. Apparently. No. I mean, he didn't kill the grandma. 776 00:50:56,710 --> 00:50:59,650 Good hackers. You know. All right. All right. They're some good coders. Trust me. 777 00:51:00,113 --> 00:51:03,324 They're not— they, they have the skills. Um, I,

778 00:51:03,405 --> 00:51:06,343 I say this because there's been photos of modded boards, 779 00:51:07,178 --> 00:51:11,529 modded Starlink boards now that have surfaced. Um, so they 780 00:51:12,029 --> 00:51:15,543 started hacking the boards to believe that they were, um, 781 00:51:15,608 --> 00:51:19,108 stationary while they were actually moving. 782 00:51:19,606 --> 00:51:22,769 Okay. Yes. Okay. Yeah. And, and this was just a standard behavior. 783 00:51:22,913 --> 00:51:26,637 I think this must have been going on for over a year because while 784 00:51:26,685 --> 00:51:29,780 it was a hack, so Ukraine just sort of was 785 00:51:30,116 --> 00:51:33,195 operating as if this was kind of standard behavior. Okay. Yeah. 786 00:51:34,686 --> 00:51:37,509 Um, and so, uh, 787 00:51:37,974 --> 00:51:41,726 what, um, Musk did 788 00:51:41,758 --> 00:51:45,142 apparently is he shut off that loophole by, 789 00:51:46,072 --> 00:51:49,824 you know, instead of tracking it on the unit, they were tracking it remotely. 790 00:51:50,000 --> 00:51:53,578 Like, if a, if a client is moving across too 791 00:51:53,610 --> 00:51:57,401 many zones too quickly. Yeah, without— you're gonna, you're gonna get— you're 792 00:51:57,901 --> 00:52:01,385 gonna get communication response— you're gonna get communication response times back from 793 00:52:01,738 --> 00:52:05,995 the communication that you're giving to the device itself. Then you can approximate 794 00:52:06,495 --> 00:52:10,638 distances, um, from that way. Even if you can't do a full triangulation 795 00:52:11,138 --> 00:52:14,316 like GPS, you could still do an approximation and get enough of a, 796 00:52:14,669 --> 00:52:18,332 a delta that you can determine that. That makes sense. Okay, I'm following, 797 00:52:18,444 --> 00:52:22,013 sir. Right, so, so basically So basically what Musk 798 00:52:22,046 --> 00:52:25,358 did is he closed the loophole, said, nope, we're not going to let 799 00:52:25,858 --> 00:52:27,257 you hack the GPS. We're going to track you server side, 800 00:52:28,831 --> 00:52:32,455 which then made their drones that they were using to hit 801 00:52:32,552 --> 00:52:34,057 Russian positions not work anymore, 802 00:52:35,784 --> 00:52:38,887 which they shouldn't have been doing in the first place. Right. Let's be clear here. 803 00:52:41,440 --> 00:52:44,937 Starlink has been credited by basically everybody with keeping 804 00:52:44,969 --> 00:52:48,445 Ukraine in the fight. And were it to be 805 00:52:48,945 --> 00:52:52,610 turned off like they claim it was, they would be— they would be 806 00:52:53,110 --> 00:52:56,857 cooked. Mhmm. Mhmm. Because it's the only communications network. 807 00:52:57,802 --> 00:53:01,183 Again, I think this is broadly accepted by, by everyone. It's the only 808 00:53:01,683 --> 00:53:05,349 communication network the Russians can't jam. So were 809 00:53:05,849 --> 00:53:09,611 it to be taken away, they would be in real trouble. They tried to push 810

00:53:10,111 --> 00:53:13,568 their use far beyond what the vendor was, was, was comfortable 811 00:53:13,681 --> 00:53:17,770 providing them. And when the vendor enforced its 812 00:53:18,270 --> 00:53:23,050 policy, they are causing an international incident, 813 00:53:23,690 --> 00:53:27,610 you know, accusing the vendor of treason effectively. Yeah. Here in the 814 00:53:28,110 --> 00:53:31,530 U.S., they're now like calling you know Musk a traitor and 815 00:53:32,030 --> 00:53:35,310 that he should, you know, as soon as Democrats come to power, they're gonna nationalize 816 00:53:35,810 --> 00:53:39,350 Starlink. We're getting there. We're getting there. We're getting there. Don't don't you 817 00:53:39,850 --> 00:53:42,770 worry. How much of this did you know and how much of this did you 818 00:53:43,270 --> 00:53:46,529 not know? The very specific details you are giving 819 00:53:46,561 --> 00:53:49,685 right now about what was being turned off, 820 00:53:50,040 --> 00:53:53,211 turned off, quote unquote, or things, that is 821 00:53:53,308 --> 00:53:56,764 what I am learning from you right now. I am learning the important 822 00:53:57,745 --> 00:54:02,073 detailed minutia of the disagreement and those 823 00:54:02,105 --> 00:54:02,959 supposed dynamics. 824 00:54:05,770 --> 00:54:09,037 And so when Zelensky says, oh, it's been turned 825 00:54:09,537 --> 00:54:13,161 back on, I don't know what he's talking about. I don't 826 00:54:13,661 --> 00:54:16,768 think he knows what he's talking about. So my question— this is all— my question 827 00:54:16,784 --> 00:54:20,358 is, why doesn't Starshield have the capability 828 00:54:20,390 --> 00:54:23,596 to do the mobile communication? Why? 829 00:54:24,269 --> 00:54:27,218 Because my— frankly, I don't know. I have no idea. Okay, 830 00:54:27,282 --> 00:54:30,199 because historically, the way that you, you outlined this, 831 00:54:30,392 --> 00:54:34,351 this, uh, dynamic, it kind of just sounds like he selected a, 832 00:54:34,880 --> 00:54:38,952 a subset of his network and then called it a new thing to 833 00:54:39,048 --> 00:54:42,292 break it off into, uh, you know, government, you know, 834 00:54:43,159 --> 00:54:46,627 DOW control. And then hypothetically, the feature set should be somewhat 835 00:54:47,127 --> 00:54:50,624 identical. So that's that in my immediate thoughts. So it's, 836 00:54:50,705 --> 00:54:54,703 it's not that clear because, for example, there are satellites 837 00:54:55,203 --> 00:54:58,331 that are only Starshield and not Starlink that are owned by the US government. 838 00:54:58,813 --> 00:55:02,714 So, and, and the capability set is different. Like, there are some capabilities 839 00:55:02,762 --> 00:55:06,232 apparently that Starshield has that Starlink doesn't Okay. It's not 840 00:55:06,457 --> 00:55:09,960 super clear, you know, how everything is working. And for all 841 00:55:10,056 --> 00:55:13,141 we know, it could be that the US government did not want them doing this, 842 00:55:13,832 --> 00:55:16,949 so would not allow them to do it on Starshield. 843 00:55:17,672 --> 00:55:20,725 And the only way they were doing it was by, you know, they found a 844 00:55:21,225 --> 00:55:24,646 way around, they found a loophole. You'll see some people talking about the closed loophole, 845 00:55:25,241 --> 00:55:27,876 which I explain now. And so they could do it on Starlink, so they didn't 846 00:55:27,892 --> 00:55:31,715 have to talk to the Pentagon at all about this. And, 847 00:55:31,763 --> 00:55:35,809 you know, what was it? Ask for forgiveness, not permission. 848 00:55:36,243 --> 00:55:40,385 In this case, it was make the others ask you for forgiveness. 849 00:55:41,381 --> 00:55:44,994 Yeah, yeah. Well, I— listen, that— you, you clarified a 850 00:55:45,494 --> 00:55:49,056 lot about what was going on. I thank you for that, um, explainer. That does 851 00:55:49,152 --> 00:55:52,428 make, uh, provided there's nothing missing in your explanation, 852 00:55:52,460 --> 00:55:56,363 I think that, that, that firmly explains Why 853 00:55:58,373 --> 00:56:01,863 not only do the news stories contain a lack of specifics 854 00:56:02,233 --> 00:56:06,727 about what is being of contention, it also explains 855 00:56:07,227 --> 00:56:11,230 how in the game of hot potato that they're playing, 856 00:56:11,278 --> 00:56:15,338 they're all able to blame each other. That explains— 857 00:56:15,451 --> 00:56:18,766 There is an additional element of stupidity here, which is that what Starlink 858 00:56:19,266 --> 00:56:23,180 then did when they said they closed the loophole is they limited the velocity 859 00:56:23,727 --> 00:56:27,113 And I'm using this word on purpose. I'll explain why. That a node can be 860 00:56:27,613 --> 00:56:30,482 moving, right? Back to down, back down to the policy level. 861 00:56:31,268 --> 00:56:34,798 Now, when people hear that they limited the speed, 862 00:56:35,360 --> 00:56:39,163 what they think is they, they added latency or they reduced the bandwidth, 863 00:56:39,548 --> 00:56:42,452 which has nothing to do with this. Yes, yes, yes, yes, 864 00:56:43,463 --> 00:56:46,736 yes. I had a— I, I was trying to explain this to my wife and 865 00:56:46,753 --> 00:56:49,994 she was getting really upset with me because I was like, sorry, when I say 866 00:56:50,494 --> 00:56:54,250 speed, are you hearing Like bandwidth, because what I mean is 867 00:56:54,750 --> 00:56:58,430 velocity, right? Yeah. Like completely different thing. Nothing to do with each other. 868 00:56:58,670 --> 00:57:02,250 Right, right, right, right. They're not, they're not throttling. They're not putting them on a 869 00:57:02,750 --> 00:57:06,370 56K modem. No, no. They're just saying, 870 00:57:07,070 --> 00:57:10,990 you know, if you're, if you, if you look like a missile, we might 871 00:57:11,490 --> 00:57:15,090 not provide you with service. Thank you. To our, 872 00:57:15,590 --> 00:57:19,399 to our

civilian, you know, service. Okay, all right. That makes that 873 00:57:19,899 --> 00:57:22,310 makes a lot of sense. That makes a lot of sense. 874 00:57:23,324 --> 00:57:27,073 Donald J. Trump, PhD, very stable genius, 875 00:57:27,510 --> 00:57:30,709 has requested to speak. So hopefully he has something very 876 00:57:30,741 --> 00:57:34,684 specific to help us illuminate some of this situation with. 877 00:57:36,799 --> 00:57:40,708 Oh yeah, there you go. Hey guys. So yeah, you're asking about the 878 00:57:40,918 --> 00:57:44,376 why Star Shield might not have the same mobile capability right now. Yes, 879 00:57:44,489 --> 00:57:46,783 sir. It's because of if I was gonna make a guess, it's because of encryption. 880 00:57:48,640 --> 00:57:52,978 requirements for military hardware. So every 881 00:57:53,316 --> 00:57:56,578 communication system has some kind of encryption attached to it, 882 00:57:56,770 --> 00:57:59,392 and that gets loaded at a certain time. 883 00:58:00,600 --> 00:58:04,134 At a certain time every week, everyone rolls crypto. So the 884 00:58:04,199 --> 00:58:07,896 problem is, if you have, let's say, 40 terminals, 885 00:58:08,172 --> 00:58:11,968 40 drones, I have to update that. 886 00:58:12,032 --> 00:58:16,578 But I also, when I'm using that drone, have to guarantee that somehow That 887 00:58:17,078 --> 00:58:20,880 communication system is completely blown when it's like basically dropped, 888 00:58:20,993 --> 00:58:25,038 when it's basically used as an explosive. Because if I can't guarantee that that 889 00:58:25,102 --> 00:58:29,259 terminal's gone, then we all have to roll the encryption systems 890 00:58:29,759 --> 00:58:33,048 that are being— the encryption keys that are being used on all the 891 00:58:33,548 --> 00:58:37,221 terminals across the board. So— Oh boy, do I have some solutions 892 00:58:37,721 --> 00:58:39,629 for you later. 893 00:58:40,512 --> 00:58:44,861 But yeah, you're in a weird situation then where You 894 00:58:45,361 --> 00:58:48,295 know, you have to have people up front that are carrying multiple keys on their 895 00:58:48,795 --> 00:58:52,225 kicks, and they have to be able to load those quickly. And, you know, 896 00:58:52,370 --> 00:58:55,611 you don't know if you're— if any of that stuff's recoverable. So I 897 00:58:56,111 --> 00:58:59,830 think that's just a problem they're having to try to figure out. And they're also 898 00:58:59,862 --> 00:59:03,376 having to try to figure out how to get the satellites to talk 899 00:59:03,876 --> 00:59:07,355 quickly enough between encryptions when they switch satellites so they don't lose, 900 00:59:07,387 --> 00:59:10,243 you know, connectivity. 901 00:59:10,419 --> 00:59:14,019 That makes conceptual sense, but if they're If their option 902 00:59:14,068 --> 00:59:17,120 is to then go to consumer, 903 00:59:18,550 --> 00:59:21,949 where there isn't any sort of cryptography to speak of 904 00:59:22,449 --> 00:59:26,143 except you know roll your own, how is that ameliorated? 905 00:59:26,191 --> 00:59:30,258 Because now you're just doing it in the clear, except for the normal encryption 906 00:59:30,758 --> 00:59:34,054 that would be on. Yeah, because you're just stuck. You're stuck between two two worlds, 907 00:59:34,087 --> 00:59:37,399 right? That's that's ultimately what it comes down to. Thank you, 908 00:59:37,822 --> 00:59:42,559 DJT. I was expecting you would. Enlighten us. Although I 909 00:59:42,656 --> 00:59:46,033 guess that could make perfect sense. I'm not complaining 910 00:59:46,065 --> 00:59:48,699 about your answer. I think your answer makes a lot of sense. I'm just trying 911 00:59:48,731 --> 00:59:50,758 to navigate the results of it in real time. 912 00:59:52,270 --> 00:59:55,599 That could make total sense as far as the US 913 00:59:55,631 --> 00:59:59,344 government authorizing certain capabilities 914 01:00:00,007 --> 01:00:03,306 because they haven't solved this sort of cryptography 915 01:00:03,338 --> 01:00:07,091 problem, or it doesn't easily provide them 916 01:00:07,140 --> 01:00:11,246 the mechanisms to solve the cryptography problem. And that 917 01:00:11,746 --> 01:00:15,043 the Ukraine is going like, well, okay, if we don't have the feature at 918 01:00:15,543 --> 01:00:18,222 all, we might as well do it in the clear and just take the risks 919 01:00:19,153 --> 01:00:22,573 that that provides, the problems 920 01:00:22,589 --> 01:00:26,395 that that provides, and deal with them as is. I could see that. Or perhaps 921 01:00:26,459 --> 01:00:30,505 they're doing some sort of throwaway encryption piecemeal 922 01:00:30,874 --> 01:00:34,246 on a case-by-case basis where they don't necessarily care because they're not going to try 923 01:00:34,262 --> 01:00:37,745 to recover the item. Yeah, but the, 924 01:00:38,018 --> 01:00:41,291 the problem is that, you know, no matter what happens, that's what 925 01:00:41,791 --> 01:00:45,286 they did. And then when they got caught, their instinct was 926 01:00:45,334 --> 01:00:47,050 to blame Trump and Elon. 927 01:00:48,206 --> 01:00:51,607 Yeah. Right, which is absurd. And, and this is why I say, 928 01:00:51,735 --> 01:00:54,848 like, in my view, Elon should never have done 929 01:00:55,348 --> 01:00:58,698 the first part of going over there playing, you know, playing hero, 930 01:00:58,714 --> 01:01:02,621 playing revolutionary, giving them all the Starlinks and and, you know, helping them, 931 01:01:02,976 --> 01:01:07,204 whatever. No good deed goes unpunished. And this 932 01:01:07,269 --> 01:01:10,642 is what he's finding himself in now, because this is no longer 933 01:01:11,142 --> 01:01:14,571 like messing with the people that like are upset with like USAID 934 01:01:15,071 --> 01:01:18,403 and Doge. He's messing with the people that are upset about Ukraine, and those people, 935 01:01:18,837 --> 01:01:19,981 uh, have a lot more pull. 936 01:01:23,190 --> 01:01:26,721 I mean, listen, out of, out of all

of his, all of his enterprises, 937 01:01:26,931 --> 01:01:30,252 Starlink seems to be the most successful and stable version of all 938 01:01:30,365 --> 01:01:33,281 of Even in my perspective, my view, 939 01:01:33,297 --> 01:01:36,759 Tesla. Starlink seems to be the 940 01:01:37,064 --> 01:01:40,607 most successful and stable business idea that 941 01:01:41,107 --> 01:01:45,047 he's had and the most impressive because it pisses 942 01:01:45,095 --> 01:01:49,135 the most people off, I think. Its success pisses 943 01:01:49,167 --> 01:01:52,838 the most people off, as we're going to get to. Yes. Yes. 944 01:01:53,030 --> 01:01:56,076 So I'll hold off on that part of the conversation. But yes, 945 01:01:56,605 --> 01:01:59,843 it is wild. So if you 946 01:01:59,907 --> 01:02:03,386 are saying that withholding some form of service For 947 01:02:03,578 --> 01:02:07,042 a few days is risking you know all these Ukrainian 948 01:02:07,542 --> 01:02:11,371 lives. You have to like by logic, you must 949 01:02:11,871 --> 01:02:15,188 credit Elon Musk for defending all those Ukrainian 950 01:02:15,220 --> 01:02:18,154 lives for years. You can't have it both ways, 951 01:02:19,148 --> 01:02:22,820 right? He can't be risking Ukrainian lives now if he 952 01:02:23,320 --> 01:02:27,310 isn't you know personally solely responsible since he was the only person in the 953 01:02:27,810 --> 01:02:31,179 world to do this and was being mocked for trying. it. When he 954 01:02:31,679 --> 01:02:34,273 pulls it off and he rides in like a fucking cowboy, 955 01:02:35,997 --> 01:02:39,688 right? And he saves your ass and he continues saving 956 01:02:39,704 --> 01:02:43,511 your ass for years, right? If you 957 01:02:44,011 --> 01:02:47,770 want to say, look, you know, I, I understand the part that 958 01:02:48,270 --> 01:02:51,038 says no, he didn't do anything special, this was good advertising for him and he 959 01:02:51,087 --> 01:02:53,693 just rode in there and he helped them out and good, and he's getting paid 960 01:02:54,193 --> 01:02:57,922 his money and he's a service provider, and that's fine. Cool. But then when 961 01:02:58,422 --> 01:03:01,445 he enforces the terms of service, you can't go and say like, "Oh my God, 962 01:03:01,461 --> 01:03:05,016 you're killing people! How dare you!" Blah blah blah blah blah. It's like, 963 01:03:05,161 --> 01:03:08,457 what are we? Are we in a moral frame or are we in a commercial 964 01:03:08,957 --> 01:03:11,278 frame? You know, you can't do both. 965 01:03:12,517 --> 01:03:15,728 Right. Anyway, go ahead. Well, Alex, let's get to. 966 01:03:17,327 --> 01:03:21,121 There's there's a group of a group of paid individuals primarily 967 01:03:21,299 --> 01:03:24,599 from the Bulwark, but also. The other, 968 01:03:25,273 --> 01:03:29,594 we'll call them popular podcasters 969 01:03:29,787 --> 01:03:33,321 in the leftist hemisphere, the left hemisphere, the Democrat, the former 970 01:03:33,821 --> 01:03:36,679 DNC hemisphere. So the people who would've been on Ms. 971 01:03:36,775 --> 01:03:40,181 Now's, this first one here is 972 01:03:40,681 --> 01:03:44,534 from Jennifer Welch, followed by The Bulwark. No, no, I have 973 01:03:44,566 --> 01:03:47,779 to make a significant correction here because these are not your average 974 01:03:48,279 --> 01:03:50,927 Dems. These are people who were formerly Republicans, 975 01:03:52,354 --> 01:03:56,680 became Never Trumpers. All of them became the Lincoln Project. The Bulwark 976 01:03:57,180 --> 01:04:00,285 has a lot of those, but not— what about Jennifer Welch? The Bulwark? 977 01:04:01,567 --> 01:04:05,861 I don't know about her. I'm sure now they've made friends with all the trans-loving 978 01:04:06,134 --> 01:04:09,355 people that they were hating before because their 979 01:04:09,855 --> 01:04:13,793 number one topic is not that. But I'm just saying, 980 01:04:14,290 --> 01:04:17,767 what's the name of the editor of the Weekly 981 01:04:18,267 --> 01:04:21,708 Standard? That's ultimately the guy behind a lot of these 982 01:04:21,724 --> 01:04:25,445 people. You're right. That that his name is escaping 983 01:04:25,945 --> 01:04:27,200 me just this moment. I will grab it. 984 01:04:29,612 --> 01:04:33,020 Bill Crystal, Bill Crystal, Bill Crystal, Bill Crystal, 985 01:04:33,133 --> 01:04:36,429 famous war hawk of Republican persuasion who has now become a Democrat and 986 01:04:36,929 --> 01:04:40,429 is aligned with the Bulwark and you know all these people and make of that 987 01:04:40,929 --> 01:04:44,265 what you will. Correct. But this first one is Jennifer Welsh, 988 01:04:44,431 --> 01:04:48,691 and she's the she's in. the MSNOW kind 989 01:04:49,191 --> 01:04:54,982 of typical person who has one of the most popular Democrat-leaning/Democrat-sided 990 01:04:55,175 --> 01:04:59,481 podcasts. And her takes are typically wild 991 01:04:59,818 --> 01:05:02,922 and inflammatory on purpose, of course, to try to get her traction. 992 01:05:02,970 --> 01:05:05,851 But here she is on Elon Musk. 993 01:05:05,964 --> 01:05:09,827 It's one man, Elon Musk, who is a white nationalist, 994 01:05:10,407 --> 01:05:14,030 who is a Nazi, who does back-to-back sick heils, who is 995 01:05:14,143 --> 01:05:15,777 an untreated drug addict, 996 01:05:16,941 --> 01:05:19,416 Spends all of his free time on X. 997 01:05:20,862 --> 01:05:24,847 He is denigrating American business and the American 998 01:05:24,879 --> 01:05:27,980 brand because now, how many decades is it going 999 01:05:28,012 --> 01:05:30,792 to take to restore the United States from this? 1000 01:05:31,306 --> 01:05:34,890 We allow one man to come into the United 1001 01:05:34,922 --> 01:05:38,023 States government, dismantle all of our soft power via the 1002 01:05:38,055 --> 01:05:41,477 USAID, and now he owns this company, 1003 01:05:41,510 --> 01:05:45,349 Starlink. that is incredibly powerful, and he will 1004 01:05:45,849 --> 01:05:49,075 tilt the

scales towards autocracy, 1005 01:05:49,364 --> 01:05:53,717 towards authoritarianism, towards war crimes against our alleged 1006 01:05:53,782 --> 01:05:57,781 ally, which Ukraine is supposed to be our ally. The James Lankfords of the world 1007 01:05:58,054 --> 01:06:01,331 are supposed to defend democracy, but they 1008 01:06:01,831 --> 01:06:05,491 don't. They do not do it. Because what do the tech oligarchs 1009 01:06:05,989 --> 01:06:09,475 and the Christian nationalists have in common? They're all 1010 01:06:09,975 --> 01:06:13,669 women. They want control. They want patriarchal control. And they're 1011 01:06:13,685 --> 01:06:17,322 willing to collude with one another in order to 1012 01:06:17,822 --> 01:06:21,312 achieve that, because they want to dismantle democracy. Because the James Lankfords 1013 01:06:21,812 --> 01:06:26,104 of the world, he's too big of a puss boy to stand up for 1014 01:06:26,120 --> 01:06:29,853 little girls that were raped by Jeffrey Epstein and 1015 01:06:29,885 --> 01:06:32,690 all of the other people. Donald Trump is named in that. 1016 01:06:33,090 --> 01:06:36,617 They're too big too big cowards to stand up to Vladimir 1017 01:06:37,117 --> 01:06:40,080 Putin, who kidnaps little Ukrainian kids and commits war crimes, 1018 01:06:40,529 --> 01:06:43,751 throws oligarchs out the window, and somehow they think they're all going to 1019 01:06:44,251 --> 01:06:48,288 be safe in this. The Republican Party could end this, but they refuse 1020 01:06:48,400 --> 01:06:52,792 to. They refuse to do it because they're all liars. James Lankford 1021 01:06:52,824 --> 01:06:56,768 is a fraud of a Christian because he values money more 1022 01:06:56,880 --> 01:07:00,572 than he does human life, and he's just one of multiple, 1023 01:07:00,765 --> 01:07:04,549 multiple Republicans that are both in government and 1024 01:07:05,049 --> 01:07:08,606 your next-door neighbors that, that lie and tell you they're big Christians 1025 01:07:08,639 --> 01:07:12,171 and they care about life, but every single time they can vote, they vote for 1026 01:07:12,203 --> 01:07:15,772 death. Now, 1027 01:07:15,788 --> 01:07:19,560 Alex, I don't know if you were aware that there's a very specific term 1028 01:07:20,060 --> 01:07:23,342 for throwing someone out of the window, and it's called defenestration, which I just don't— 1029 01:07:23,474 --> 01:07:27,728 as a word, I don't know how you would ever from context 1030 01:07:28,228 --> 01:07:31,606 clues interpret that as being the process of throwing someone out 1031 01:07:32,106 --> 01:07:36,044 of a window. But just for the record, defenestration is the act of throwing someone 1032 01:07:36,060 --> 01:07:39,954 out of a window. So I just wanted to point that out. It doesn't 1033 01:07:40,454 --> 01:07:43,031 sound so bad though. It doesn't. It's not quite as graphic. 1034 01:07:44,168 --> 01:07:49,023 Correct. Correct. Now, Alex, before we move on, do you have any specific responses 1035 01:07:49,055 --> 01:07:52,100 to that wild sort of fill-in-the-blanks 1036 01:07:52,600 --> 01:07:54,758 Mad Libs sort of Mad Lib, 1037 01:07:56,431 --> 01:07:59,987 right? I mean, no, because look, 1038 01:08:00,212 --> 01:08:03,501 no, I actually, I do. Um, here's the thing, right? We've been, 1039 01:08:04,180 --> 01:08:07,385 we and friends of ours have been called conspiracy theorists, I think, from morning till 1040 01:08:07,498 --> 01:08:10,168 night for a few years throughout COVID. 1041 01:08:10,799 --> 01:08:14,070 Is that, is that fair? I, I think yes. You know, there have been breaks, 1042 01:08:14,559 --> 01:08:19,170 but it's, it's been, you know, excruciatingly common and 1043 01:08:20,020 --> 01:08:23,055 You know, your first response is like, wait, like, 1044 01:08:23,152 --> 01:08:26,348 you don't think people coordinate? Like, is Signal, 1045 01:08:26,621 --> 01:08:29,801 you know, chat rooms is weird to you? 1046 01:08:29,993 --> 01:08:34,329 Like, you know, when we say these guys are working together to 1047 01:08:34,829 --> 01:08:38,746 accomplish an objective, like, you know, I, I text my wife, let's take the kids 1048 01:08:39,246 --> 01:08:42,761 to the park. Like, we don't announce this, right? Like, it's— but it's fair to 1049 01:08:43,261 --> 01:08:47,531 assume that we are coordinating because we're on the same 1050 01:08:48,031 --> 01:08:51,548 Same team here. Now remember, Sam Harris says, well, Sam Harris says, is it conspiracy 1051 01:08:51,676 --> 01:08:55,201 if it's conspiracy out in the open? That's right. That's right. 1052 01:08:55,281 --> 01:08:57,989 Exactly. So, you know, I mean, is it, you know, whatever? 1053 01:08:58,790 --> 01:09:01,738 No. Yes. Yes. If it's about an asteroid, 1054 01:09:02,844 --> 01:09:04,366 hey, that eats an asteroid. 1055 01:09:05,840 --> 01:09:09,237 Tiny asteroids. Or forcing a kid to eat peanut butter, is it? 1056 01:09:10,311 --> 01:09:13,739 Anyway, sorry. We can do this forever. But the point is. 1057 01:09:15,440 --> 01:09:19,164 You know what I've come to realize, probably too 1058 01:09:19,664 --> 01:09:23,081 late, is that they they have no problem with conspiracy theories, right? What she just 1059 01:09:23,581 --> 01:09:27,046 like that tirade there, I connected Musk and Trump and Epstein filed and 1060 01:09:27,175 --> 01:09:31,413 the Republicans and you know Ukraine and I don't 1061 01:09:31,589 --> 01:09:33,901 need I you know the list is partial, 1062 01:09:35,185 --> 01:09:38,011 right? And it's he's a Nazi and he's a white supremacist. Yeah, 1063 01:09:38,476 --> 01:09:42,601 okay. All of these connections which somehow lead to him. of— and then 1064 01:09:43,101 --> 01:09:45,875 he controls his company. How did he control it? Maybe he created it? No, 1065 01:09:46,261 --> 01:09:49,053 no credit to it. Fine. Okay. Um, 1066 01:09:50,048 --> 01:09:54,558 like, they, they have zero problem conspiratizing, but their 1067 01:09:54,622 --> 01:09:58,073 problem is

conspiracies that are aimed at them, 1068 01:09:58,619 --> 01:10:01,765 right? Like, because that is a hostile act, you know, 1069 01:10:01,989 --> 01:10:05,055 conspiratizing against, you know, one group of 1070 01:10:05,555 --> 01:10:09,326 people who happen to have the power. Is you 1071 01:10:09,826 --> 01:10:13,967 know heavily sort of guarded against because that is a weapon, 1072 01:10:14,208 --> 01:10:17,565 right? That separates one group of people from another inside the 1073 01:10:18,065 --> 01:10:21,515 same society, and they I think correctly see it as a weapon of information 1074 01:10:22,015 --> 01:10:24,743 warfare. But by the same token, 1075 01:10:25,048 --> 01:10:28,373 you know, like the state sort of claims the monopoly on violence, 1076 01:10:28,903 --> 01:10:32,885 they also claim the monopoly on conspiracy theorizing because 1077 01:10:32,998 --> 01:10:35,069 you know when when they want to do it, right? When they want to go 1078 01:10:35,166 --> 01:10:38,957 on for years and years about Russiagate. This and that, and Kalimnik 1079 01:10:39,457 --> 01:10:43,198 and this, you know, whatever, like random Russian-named 1080 01:10:43,698 --> 01:10:47,214 people. Um, it's totally fine. There's no problem. 1081 01:10:47,552 --> 01:10:51,199 Like, nobody loses any standing in the elites for, for doing that. 1082 01:10:52,259 --> 01:10:55,858 Um, it's just that when the conspiritizing is, 1083 01:10:56,050 --> 01:11:00,147 is turned against them, correctly or incorrectly, right? There are some pretty batshit 1084 01:11:00,372 --> 01:11:03,553 conspiracy theories out there. Doesn't mean just because it's conspiracy theory is correct. 1085 01:11:04,323 --> 01:11:07,226 But when it's turned against them, that's when they have a problem with it. 1086 01:11:07,692 --> 01:11:11,397 The quality of the conspiritizing is irrelevant. It's the target 1087 01:11:11,461 --> 01:11:15,310 that they object to. There's a— not to belabor this point 1088 01:11:15,375 --> 01:11:19,064 too long, but there's an effect that exists in that 1089 01:11:19,401 --> 01:11:23,138 dynamic that gives them that sort of 1090 01:11:23,170 --> 01:11:26,346 hubris, where if one 1091 01:11:26,458 --> 01:11:30,259 person who occupies a category they 1092 01:11:30,372 --> 01:11:31,799 get to claim is respectable, 1093 01:11:33,116 --> 01:11:37,081 Respectable opinion, that person is allowed to float 1094 01:11:37,129 --> 01:11:40,308 the conspiracy theory, then give them all authorization 1095 01:11:40,404 --> 01:11:44,161 to repeat the conspiracy theory because in their minds, it's not insane 1096 01:11:44,337 --> 01:11:47,564 if all of their friends or this person in particular is the first 1097 01:11:48,064 --> 01:11:51,562 person to edit it. Reasonable people have thought is the 1098 01:11:52,062 --> 01:11:55,479 way that they sort of short circuit that sort of idea in their heads. 1099 01:11:55,575 --> 01:11:59,830 Reasonable people have said, reasonable people have suspected. did. Reasonable people, 1100 01:11:59,846 --> 01:12:03,042 blah blah blah, reasonable people fill in the blanks. That is how they 1101 01:12:03,542 --> 01:12:06,847 allow themselves to short-circuit that, that sort of phenomenon. So when you have— 1102 01:12:07,347 --> 01:12:11,087 If reasonable people have said it, then the thing is reasonable. It's obvious. 1103 01:12:11,713 --> 01:12:15,262 Exactly, exactly. It's a reasonable thing to say because a reasonable person said it. 1104 01:12:15,567 --> 01:12:18,795 That's logic. There you go. So that's, that's how they get around it. It doesn't 1105 01:12:19,295 --> 01:12:22,215 matter how batshit insane the thing is. It doesn't matter if you spend 3 years 1106 01:12:22,713 --> 01:12:26,214 on a Russian, uh, conspiracy that never pans out and, and spend 1107 01:12:26,714 --> 01:12:30,031 spend $30 billion of government money having an 1108 01:12:30,209 --> 01:12:33,897 elderly guy who is going 1109 01:12:33,929 --> 01:12:37,527 senile go through Mueller and forget where he is and 1110 01:12:38,043 --> 01:12:41,361 who he's talked to or anything like that. It doesn't matter if you spend 1111 01:12:41,425 --> 01:12:45,225 all that time. I mean, listen, I've played the Rachel Maddow clip 1112 01:12:45,546 --> 01:12:49,322 where she has the former 1113 01:12:49,822 --> 01:12:53,790 shyster attorney come on and start to accuse 1114 01:12:55,109 --> 01:12:58,367 Brett Kavanaugh of having run extensive gang rape 1115 01:12:58,640 --> 01:13:01,352 campaigns in college on women. 1116 01:13:01,689 --> 01:13:04,818 They're running trains on people, as they said. 1117 01:13:06,134 --> 01:13:09,760 It doesn't matter how crazy the idea is. If one 1118 01:13:09,889 --> 01:13:13,307 person who is in their respectable class throws the idea 1119 01:13:13,403 --> 01:13:16,949 out there, it's all fair. So now moving forward real quick to the 1120 01:13:17,449 --> 01:13:20,913 remaining here, these are the Bulwark. This is Kara Swisher being interviewed by 1121 01:13:21,413 --> 01:13:25,578 the Bulwark. Bulwark, followed by J.V. last at some sort of Bulwark convention 1122 01:13:26,251 --> 01:13:29,650 situation. Under Kamala Harris, he might be in jail right now. 1123 01:13:30,163 --> 01:13:34,043 He is correct. And that's not because she's some avenging 1124 01:13:34,107 --> 01:13:38,082 fury. It's because there's a lot of questions around different companies 1125 01:13:38,179 --> 01:13:41,144 of his. If Donald Trump loses the House and the Senate, 1126 01:13:41,305 --> 01:13:44,511 I would worry. I'd be worried if I were him. No matter 1127 01:13:45,011 --> 01:13:47,269 how rich he is, there were all these agencies, some of which he put a 1128 01:13:47,769 --> 01:13:51,461 knife in and killed in the DOJ, many of whom were investigating him. He stopped 1129 01:13:51,493 --> 01:13:55,170 like half a dozen investigations of his various companies. So they're 1130 01:13:55,670 --> 01:13:58,013 going to

start investigating again. You think he's not on the list? He's at the 1131 01:13:58,513 --> 01:14:00,871 top of that list. Donald Trump obviously is at the top and his family, 1132 01:14:00,903 --> 01:14:04,821 but Elon Musk is right up there for that. And you 1133 01:14:05,321 --> 01:14:08,660 have to— he knows what's coming. Like, he's not stupid. 1134 01:14:08,756 --> 01:14:12,674 Like, of all the many things you can say about him is he, he has 1135 01:14:12,771 --> 01:14:16,679 a very big sense of Fate. And I think it's 1136 01:14:16,743 --> 01:14:20,199 coming for him and real hard. However mad you are 1137 01:14:20,699 --> 01:14:24,265 about Elon Musk, it's not enough. And he's the one I want. And I 1138 01:14:24,378 --> 01:14:28,058 want being hauled before Congress to be just the start, because this 1139 01:14:28,106 --> 01:14:31,771 guy has a body count like a fucking war criminal. 1140 01:14:33,339 --> 01:14:36,875 We start with the US Congress and it ends at the fucking 1141 01:14:36,972 --> 01:14:38,514 Hague for Elon Musk. 1142 01:14:41,790 --> 01:14:44,436 Hey, you got a trillion dollars, asshole? 1143 01:14:44,872 --> 01:14:48,245 Great, you start spending it on lawyers because justice is 1144 01:14:48,342 --> 01:14:51,455 coming. So, uh, we gotta talk a little bit about Elon 1145 01:14:51,520 --> 01:14:55,892 Musk. Uh, we talked— oh, 1146 01:14:55,988 --> 01:14:59,187 you don't hate him nearly enough. There has been a series of meetings 1147 01:14:59,460 --> 01:15:02,378 among House Democrats. Do you know what they're doing? Yes. 1148 01:15:03,335 --> 01:15:06,893 They're carving up jurisdiction. for who 1149 01:15:07,118 --> 01:15:10,862 is gonna ask him— sorry, ask him— subpoena 1150 01:15:10,878 --> 01:15:14,526 him to testify before which of 1151 01:15:14,591 --> 01:15:17,302 them for what. Get a townhouse, big guy. 1152 01:15:18,245 --> 01:15:21,528 Got a lot of nice brownstones right there on Capitol Hill. 1153 01:15:22,733 --> 01:15:24,847 Elon, you motherfucker, you better pack a lunch. 1154 01:15:26,860 --> 01:15:30,721 'Cause your ass is gonna be living in the Capitol starting January 1155 01:15:31,221 --> 01:15:35,022 of 2024. I feel like my guy Robert Garcia and Jamie Raskin 1156 01:15:35,607 --> 01:15:38,254 are fucking ready. They're loaded for bear. 1157 01:15:39,690 --> 01:15:42,061 What's going to happen on the policy side? I don't know about that part, 1158 01:15:42,142 --> 01:15:44,708 but I think Jamie and Robert are going to be ready, loaded for bear. 1159 01:15:45,309 --> 01:15:49,105 And I just, I hope that Elon and Jared, 1160 01:15:49,364 --> 01:15:52,050 you know, should, they should start texting about property right now. 1161 01:15:52,374 --> 01:15:53,879 I think they should get a real estate agent. 1162 01:15:56,639 --> 01:16:00,487 The Hague, Alex. The Hague. Now, what happens 1163 01:16:00,535 --> 01:16:04,136 at The Hague, classically, if you bring someone 1164 01:16:04,314 --> 01:16:06,933 up there for— Well, I mean, 1165 01:16:07,078 --> 01:16:09,362 Milošević famously died there. 1166 01:16:10,456 --> 01:16:12,982 Why? What happened? What happened? How did he die? 1167 01:16:14,832 --> 01:16:18,152 He was extradited by his own country, and I believe 1168 01:16:18,652 --> 01:16:22,031 he died waiting for trial. For war crimes, of course. 1169 01:16:22,214 --> 01:16:25,836 That's the, that's the, uh, The punchline here, 1170 01:16:26,126 --> 01:16:29,415 I guess. Look, I will say there is a 1171 01:16:29,431 --> 01:16:32,785 real problem here, right? Because Elon himself, 1172 01:16:32,897 --> 01:16:36,013 like the biological entity, is becoming 1173 01:16:36,207 --> 01:16:41,278 a nation-state actor. Like he personally 1174 01:16:41,294 --> 01:16:44,168 has capabilities that most nations don't have, 1175 01:16:45,072 --> 01:16:49,010 right? And as such, he's becoming a geopolitical entity 1176 01:16:49,510 --> 01:16:52,818 that others need to have relations with. Right? He's got a good 1177 01:16:53,318 --> 01:16:56,468 relation with China. He's got a so-so relation with Russia, but not terrible. 1178 01:16:58,265 --> 01:17:01,840 Um, you know, he's trying to keep his balances while being a U.S., 1179 01:17:02,539 --> 01:17:05,879 um, corporation CEO, multiple who which 1180 01:17:06,379 --> 01:17:10,665 are merging now. So even his own company is going to be probably the 1181 01:17:11,165 --> 01:17:14,792 biggest company in the U.S. Um, and that's 1182 01:17:15,292 --> 01:17:18,846 just going to cause problems. Right? On the one hand, how are you 1183 01:17:18,894 --> 01:17:22,115 going to bring up a CEO for war crimes? He's not a general. He doesn't 1184 01:17:22,615 --> 01:17:25,244 command any armies. On the other hand, I mean, 1185 01:17:26,887 --> 01:17:30,011 you can see how they can construct that, as crazy as it is, 1186 01:17:30,928 --> 01:17:33,762 because he has nation-state type capabilities. 1187 01:17:36,981 --> 01:17:41,525 Now, the things that slip through there, Alex, of course, is that they equate 1188 01:17:42,156 --> 01:17:46,102 the purchase that the government— the purchases that the government does for 1189 01:17:46,166 --> 01:17:49,883 his technology, uh, with some sort of government handout, 1190 01:17:50,572 --> 01:17:53,921 that they, they believe that anything that the government does that's 1191 01:17:53,937 --> 01:17:57,558 business-wise is a gift to anyone that the government does business with. 1192 01:17:58,487 --> 01:18:01,724 As opposed to— it's a sort of gift. Um, it's very funny, some of the 1193 01:18:02,224 --> 01:18:05,345 earliest contracts that SpaceX got out of the US government, they actually had to sue 1194 01:18:05,409 --> 01:18:09,784 the Department of Defense at the time, um, because they were giving no-bid 1195 01:18:10,284 --> 01:18:13,732 contracts to ULA Um, and SpaceX was like, 1196 01:18:13,748 --> 01:18:17,555 look, if we lose a fair bid, we lose a fair bid. But this is— 1197

01:18:18,037 --> 01:18:21,956 no, we're not even given a chance to compete. And we, 1198 01:18:22,326 --> 01:18:25,281 you know, they had to go to court to get a chance to compete. 1199 01:18:25,442 --> 01:18:29,201 And then they did win the contract, actually. Mm-hmm. Right? So this is, you know, 1200 01:18:29,217 --> 01:18:32,622 when they say like, oh, we paid for this. It's like, you better believe you 1201 01:18:32,767 --> 01:18:36,236 paid for it. You were forced to by a court. What are you talking 1202 01:18:36,736 --> 01:18:40,390 about? Stop wasting the money of the US taxpayer and get a better bid. 1203 01:18:41,272 --> 01:18:45,024 Like, you literally—they literally have to go to court. Like that's you know 1204 01:18:45,524 --> 01:18:48,745 when they say like you know when people hear this, they imagine that like you 1205 01:18:49,245 --> 01:18:52,080 know Elon was went to a smoke-filled room and they're like yeah we're gonna we're 1206 01:18:52,580 --> 01:18:54,726 gonna get you sorted man. Don't worry. Like yeah you just go out there and 1207 01:18:54,742 --> 01:18:58,238 we're just we're gonna make sure you're gonna make so much money whatever. They have 1208 01:18:58,414 --> 01:19:02,263 no idea how much trouble SpaceX had 1209 01:19:02,763 --> 01:19:06,115 to deal with in the early days. Including, you know, 1210 01:19:06,853 --> 01:19:10,546 its 4th failure to launch originally was because 1211 01:19:11,046 --> 01:19:14,079 they were launching from the middle of the Pacific and some of the 1212 01:19:14,579 --> 01:19:18,237 machinery, like, literally was corroded by salt, right? 1213 01:19:18,317 --> 01:19:21,737 Had they been allowed to use Vandenberg in California, they would not 1214 01:19:21,753 --> 01:19:24,194 have— like, they would have succeeded one flight earlier. 1215 01:19:25,623 --> 01:19:28,850 And part of the problem is the Air Force wasn't trusting them. And so they 1216 01:19:29,107 --> 01:19:31,885 had to, like, get some random island in the middle of the Pacific to launch 1217 01:19:32,385 --> 01:19:35,774 from. So, you know, it's easy to say 1218 01:19:35,838 --> 01:19:38,251 now in retrospect, oh, you know, we gave you all this money and you owe 1219 01:19:38,751 --> 01:19:41,861 us, motherfucker, or whatever. But in practice, like, 1220 01:19:41,877 --> 01:19:44,801 it has nothing to do— like, the, the actual facts of the situation are just 1221 01:19:45,301 --> 01:19:49,045 so different that, you know, like, it's really 1222 01:19:49,077 --> 01:19:52,354 infuriating because, like, frankly, I don't know what Elon is 1223 01:19:52,387 --> 01:19:55,021 thinking with the way he is, you know, um, 1224 01:19:55,134 --> 01:19:58,769 developing his brand on social media. Frankly, like, I don't I don't agree with 1225 01:19:59,269 --> 01:20:01,916 100% of what he says or whatever. And like, I shouldn't even have to say 1226 01:20:02,416 --> 01:20:06,002 this. It doesn't, it doesn't matter. Um, but I 1227 01:20:06,502 --> 01:20:10,255 mean, the way these people are approaching this whole situation 1228 01:20:10,755 --> 01:20:14,579 just makes it really, really like infuriating to not defend him, 1229 01:20:14,611 --> 01:20:18,569 even though again, I'm, you know, I'm not as like, 1230 01:20:20,330 --> 01:20:23,518 you know, positive on the man. Um, but I mean, you know, 1231 01:20:23,583 --> 01:20:26,998 if this is the alternative, fuck, man, like Elon for president, change the Constitution. 1232 01:20:27,046 --> 01:20:27,759 It doesn't matter. 1233 01:20:32,411 --> 01:20:35,785 Yeah, well, that's where I am. 1234 01:20:36,060 --> 01:20:40,838 In a possible potential future, can you imagine some of these people specifically 1235 01:20:41,127 --> 01:20:44,469 refusing to do business with Elon Musk just because it's Elon 1236 01:20:44,517 --> 01:20:47,908 Musk and then engaging in contracts with a company like, 1237 01:20:47,956 --> 01:20:51,509 let's say, Boeing, who has had an abysmal track 1238 01:20:52,009 --> 01:20:54,405 record on its deliverables the last decade? 1239 01:20:55,998 --> 01:20:59,472 Boeing has not had any successes that I'm aware of in the last 1240 01:20:59,972 --> 01:21:03,901 decade when it comes to technology, not only advancing technology, but having technology 1241 01:21:03,965 --> 01:21:07,367 that meets normal standards, right? Like the safety 1242 01:21:07,431 --> 01:21:10,705 standards, the functionality standards. Their stuff is not 1243 01:21:11,205 --> 01:21:14,781 working, right? Boeing's stuff is not working. When my son 1244 01:21:15,214 --> 01:21:18,696 was maybe like 1 year old or something, right? So this must have been 1245 01:21:18,744 --> 01:21:23,767 6 years ago. We watched the first human mission 1246 01:21:23,799 --> 01:21:26,928 to the ISS. from SpaceX, which was part 1247 01:21:27,428 --> 01:21:30,571 of the contract that it had secured with Boeing. So Boeing got— Right. 1248 01:21:31,021 --> 01:21:34,631 More money than SpaceX. If we believe Boeing got $4.2 billion 1249 01:21:35,131 --> 01:21:37,600 to do the same thing that SpaceX got $2.8 billion to do, okay? 1250 01:21:38,419 --> 01:21:41,757 SpaceX did accomplish the thing 6 years ago. 1251 01:21:42,190 --> 01:21:45,464 I'm sorry, SpaceX. Did I say Boeing? I meant SpaceX accomplished the 1252 01:21:45,964 --> 01:21:50,327 thing 6 years ago and has ever since been the sole provider of rides 1253 01:21:50,599 --> 01:21:54,150 to the ISS. Okay? Mhmm. Boeing still 1254 01:21:54,650 --> 01:21:58,350 has not— we're not talking about matching Starlink. 1255 01:21:58,850 --> 01:22:02,470 Forget that. Right. Right. We're talking about doing a thing that they were paid for, 1256 01:22:02,970 --> 01:22:05,680 like, I don't know, over a decade ago now at this point, 1257 01:22:06,310 --> 01:22:09,690 which they're still delaying, still failing at, still being given more 1258 01:22:10,190 --> 01:22:13,370 money because the Pentagon is, like, so fucked.

They're like, look, we can't have a 1259 01:22:13,870 --> 01:22:16,530 single provider. Give them what they need. Give them extra money. 1260 01:22:16,910 --> 01:22:20,763 Imagine, as I was heading towards, imagine, imagine a future where they 1261 01:22:21,263 --> 01:22:24,451 are specifically blocking Elon Musk out just 1262 01:22:24,951 --> 01:22:28,989 because of their personal animosity towards him, start pressuring 1263 01:22:29,021 --> 01:22:33,479 the internal apparatus to engage in contracts with companies 1264 01:22:33,527 --> 01:22:37,311 that have a lesser quality deliverable set or 1265 01:22:37,536 --> 01:22:40,935 delayed, and they just don't count those bodies, 1266 01:22:41,047 --> 01:22:45,056 quote unquote, the number of lives that wind up becoming lost in the 1267 01:22:45,556 --> 01:22:47,685 mix as they secure those things. Right. 1268 01:22:49,005 --> 01:22:52,187 Those deliverables at a minimum get pushed back multiple years. 1269 01:22:52,620 --> 01:22:56,316 They're not going to count the lives that they've cost just with their pettiness right 1270 01:22:56,816 --> 01:23:00,329 there. So whatever they want to tally up as belonging to Elon Musk, 1271 01:23:00,345 --> 01:23:03,912 they're not going to be doing that in a fair way with 1272 01:23:04,154 --> 01:23:07,679 their alternate decisions. They're going to just invent numbers 1273 01:23:07,795 --> 01:23:09,254 to hate Elon Musk regardless. 1274 01:23:11,109 --> 01:23:14,411 The thing is that you have noticed a number of 1275 01:23:14,443 --> 01:23:17,600 people talking about nationalizing SpaceX. I don't know if that's next on the list. 1276 01:23:19,858 --> 01:23:23,933 It is not specifically on the list, but of course that's always out there in 1277 01:23:24,433 --> 01:23:28,377 the, uh, you know, uh, Bernie Sanders agenda. You know why 1278 01:23:28,425 --> 01:23:31,553 though, right? Because at some point, even the 1279 01:23:32,053 --> 01:23:35,532 most knuckle-dragging of morons realize that there is 1280 01:23:35,580 --> 01:23:39,158 no alternative, man. Like, when, you know, some idiots are going with a 1281 01:23:39,658 --> 01:23:43,425 take of, um, oh, you know, well, you know, we should develop alternatives. Europe should 1282 01:23:43,925 --> 01:23:47,919 build its own Starlink. It's like, Okay, and then 15 years from 1283 01:23:48,419 --> 01:23:51,450 now when it's actually able to be where he is now, 1284 01:23:51,675 --> 01:23:55,463 then fine. To start a committee to develop a funding 1285 01:23:55,963 --> 01:24:00,167 grant to provide the resources for the right university. 1286 01:24:00,504 --> 01:24:03,891 I have my criticisms of Elon Musk, we know that. And of course you 1287 01:24:04,391 --> 01:24:08,257 can imagine I'm mocking his recent Roadster 1288 01:24:08,771 --> 01:24:12,704 demo. Okay, but the man took 1289 01:24:13,041 --> 01:24:16,590 a It was a future leap on the Starlink 1290 01:24:16,814 --> 01:24:20,315 setups. He combined his goals.

1291 01:24:20,829 --> 01:24:24,136 He developed SpaceX while trying to deliver satellites into space, 1292 01:24:24,200 --> 01:24:27,653 as well. And he got a head start on something that didn't have 1293 01:24:28,153 --> 01:24:31,153 a market. He created a market. I mean, it had a market in internet services, 1294 01:24:31,265 --> 01:24:35,103 but you didn't have a market for space internet services. There was no 1295 01:24:35,603 --> 01:24:38,218 satellite internet services that were massive enough to— This was another situation. 1296 01:24:39,120 --> 01:24:42,951 This is another situation where the U.S. was given giving like 1297 01:24:43,161 --> 01:24:46,831 ridiculous contracts to various ISPs to take fiber to like 1298 01:24:47,331 --> 01:24:50,456 some random you know ranches in the middle of nowhere, and Elon shows up 1299 01:24:50,956 --> 01:24:53,670 and is like fixed. You don't need like I'm not asking you for money. 1300 01:24:53,718 --> 01:24:57,205 I'm just telling you stop paying them because it's solved. 1301 01:24:59,622 --> 01:25:03,426 Right? But but think about it this way: If you take 1302 01:25:03,926 --> 01:25:07,585 the United States and you subtract SpaceX. Right now, right? U.S. and China 1303 01:25:08,085 --> 01:25:11,733 are neck and neck in space technology. If you add 1304 01:25:12,233 --> 01:25:14,299 SpaceX, the United States is a decade ahead. 1305 01:25:16,059 --> 01:25:19,666 Okay? The so the the Chinese just barely just just right 1306 01:25:20,166 --> 01:25:23,305 now this year managed to land a booster, 1307 01:25:23,853 --> 01:25:26,408 right? They still have to figure out how to refly it, how to refurbish it, 1308 01:25:26,424 --> 01:25:29,917 how to make that a reliable program, and then get it 1309 01:25:30,417 --> 01:25:33,630 up to. Starship levels and whatnot. 1310 01:25:34,323 --> 01:25:38,223 But they, you know, they're not jokers, right? Like, they move fast 1311 01:25:38,723 --> 01:25:41,269 and they know what they need to do. So it's probably not 10 years that 1312 01:25:41,769 --> 01:25:44,915 they need to get to where Elon is today. If you 1313 01:25:44,964 --> 01:25:48,611 subtract SpaceX from the capabilities of the United 1314 01:25:49,111 --> 01:25:52,805 States, like, you know, in the 1315 01:25:53,305 --> 01:25:55,983 one area that the United States has a clear advantage, this and AI, fine. 1316 01:25:56,515 --> 01:25:59,856 But like, in space, it has an even bigger advantage. Right. In AI, 1317 01:25:59,904 --> 01:26:03,286 like, the US is like maybe a year ahead, maybe less, 6 months, 9 months, 1318 01:26:03,334 --> 01:26:06,765 whatever. In, um, in space, 1319 01:26:07,358 --> 01:26:10,933 you've got a very clear advantage. And what they're saying is 1320 01:26:10,949 --> 01:26:14,363 like, yeah, let's, you know, the guy that did that, just take him out, 1321 01:26:15,533 --> 01:26:19,028 you know, lock him up for, for war crimes. And, 1322 01:26:20,262 --> 01:26:23,693 you know, yeah,

yeah. And there's, there's a very, there's a very— Because 1323 01:26:24,193 --> 01:26:27,688 the government should run it better, obviously. Well, there's a very perverse pathology that really 1324 01:26:27,720 --> 01:26:31,253 shows itself in moments like this, which is the fact that Elon Musk 1325 01:26:31,753 --> 01:26:34,255 took all of the risk. And now, of course, 1326 01:26:35,331 --> 01:26:39,024 when it comes to finances and stuff like that, he's got enough lawyers 1327 01:26:39,524 --> 01:26:43,359 who can figure out how to make sure that he doesn't necessarily incur any significant 1328 01:26:43,921 --> 01:26:48,272 losses specifically that hinder him in general if the project 1329 01:26:48,772 --> 01:26:50,713 goes awry. It's not like he's going to be broke and homeless. 1330 01:26:52,126 --> 01:26:55,353 But broadly speaking, he's the one who took the initiative, 1331 01:26:55,562 --> 01:26:59,158 carved out a market in something that hadn't existed yet, which is 1332 01:27:00,154 --> 01:27:03,445 the space internet facilities here, space internet 1333 01:27:03,945 --> 01:27:08,005 industry, and had made that so successful that now people 1334 01:27:08,021 --> 01:27:11,345 have decided he has too much power because of this thing 1335 01:27:11,393 --> 01:27:15,246 that he created. This business that he started has given 1336 01:27:15,746 --> 01:27:18,887 him too much power. That's right. And they want to take it from 1337 01:27:18,935 --> 01:27:22,460 him because they don't like how he uses it when 1338 01:27:22,960 --> 01:27:25,713 they, they, none of these people could have ever done this. And remember, the, 1339 01:27:25,729 --> 01:27:28,726 the, the previous take was that he's bad at business. He sucks and he's never, 1340 01:27:28,774 --> 01:27:32,252 he's gonna fail at everything. He's gonna go bankrupt, right? We have transitioned now 1341 01:27:32,412 --> 01:27:35,408 from, from that take to the take of— I mean, 1342 01:27:35,440 --> 01:27:37,956 listen, there's still plenty of places where he could, he could, he could fail, 1343 01:27:38,149 --> 01:27:41,658 but you know that, that when you at least have one major success, that's fine. 1344 01:27:42,732 --> 01:27:46,151 You know, you don't have to hit 100, you just have to hit something, 1345 01:27:46,728 --> 01:27:50,436 you know. Well, all I'm saying is like, you know, we never went 1346 01:27:50,936 --> 01:27:53,742 to the, the part where we're like, you know, he's great. It's, it's, it's like 1347 01:27:53,870 --> 01:27:57,626 either he's a shyster and a fraud, uh, or he is 1348 01:27:57,963 --> 01:28:01,044 too good. He's never good enough, 1349 01:28:01,574 --> 01:28:04,527 right? Yeah, we didn't have that, that phase, 1350 01:28:05,121 --> 01:28:08,427 you know. It's, it's, I mean, it just tells you that 1351 01:28:08,927 --> 01:28:12,421 these people People that have no understanding of making things, what it takes 1352 01:28:12,921 --> 01:28:16,432 to create novel technologies, just shoot for the moon 1353 01:28:16,770 --> 01:28:20,492 and occasionally hit it. It's not a model 1354 01:28:20,992 --> 01:28:24,197 of the world that they comprehend. No, no, not at all. 1355 01:28:24,439 --> 01:28:27,902 We have a request to speak from KK, request to 1356 01:28:28,402 --> 01:28:30,238 comment probably on this topic. Go ahead, KK. 1357 01:28:31,751 --> 01:28:34,666 Yeah, it is on this topic. I have a question for you, Alex. Are you 1358 01:28:35,166 --> 01:28:38,652 familiar with Project Meridian? Do you hear me, by the way? 1359 01:28:38,732 --> 01:28:42,088 I'm sitting outside. Yes. Okay, because I was like between 1360 01:28:42,184 --> 01:28:45,909 toggling between Wi-Fi and not. I'm putting Blair Witch Project sticks together in 1361 01:28:46,409 --> 01:28:50,710 a front yard. But anyways, Project Meridian, like, Pete Hegseth 1362 01:28:50,775 --> 01:28:54,484 announced it, was it last month at the State of the Force at Quantico, 1363 01:28:54,708 --> 01:28:58,305 which is like a 120-day effort, which is gonna be co-led by Elon. 1364 01:28:58,739 --> 01:29:02,271 And the outcome of that is gonna be the trajectory of the future of warfare. 1365 01:29:02,817 --> 01:29:06,572 Including the subterranean, you know, space domain stuff. So I wanted to know what 1366 01:29:06,620 --> 01:29:10,006 your thoughts are on his involvement and how appropriate or how 1367 01:29:10,055 --> 01:29:13,328 it should go. You can even replace Elon with someone else. When, like, 1368 01:29:14,002 --> 01:29:17,341 somebody in tech has that much aid in the future 1369 01:29:17,841 --> 01:29:21,240 of warfare with a private company, how should that be handled? 1370 01:29:21,353 --> 01:29:24,514 I mean, consulting-wise, that's how government works, 1371 01:29:24,659 --> 01:29:28,173 but we're— okay, I'll just let you 1372 01:29:28,189 --> 01:29:31,768 respond if you— So aren't people from Lockheed and Raytheon 1373 01:29:31,816 --> 01:29:34,289 and whatever have been in these roles before? 1374 01:29:35,477 --> 01:29:38,720 Yeah, well, actually, MITRE— do you know the company MITRE? No. 1375 01:29:39,250 --> 01:29:42,783 They're funding the research for that project, and there's 1376 01:29:43,283 --> 01:29:46,733 a couple other people involved, but— Right. So here's the 1377 01:29:47,233 --> 01:29:50,891 thing, right? If the US wants to win, if the 1378 01:29:51,391 --> 01:29:55,612 US conceives of itself as a power in the world that 1379 01:29:55,628 --> 01:29:59,063 has been on the top, For a number of 1380 01:29:59,095 --> 01:30:02,063 decades has some real competition now for the first time, 1381 01:30:02,609 --> 01:30:05,288 and it desires to win. It desires to stay on top, 1382 01:30:06,780 --> 01:30:10,293 and and it's realizing secondly, it's realizing that it's behind 1383 01:30:10,470 --> 01:30:13,213 in its you know in the thing that it's putting a trillion and a half 1384 01:30:13,713 --> 01:30:17,208 per year weapons

procurement. Broadly speaking, 1385 01:30:17,224 --> 01:30:20,369 okay, the the Pentagon has a bigger budget than just the weapons, 1386 01:30:20,433 --> 01:30:23,529 but a lot of it goes to weapons. They're behind. They're getting humiliated. 1387 01:30:24,267 --> 01:30:27,891 In Ukraine, in Iran, in whatever, because a 1388 01:30:27,971 --> 01:30:32,269 bunch of adversaries have been very pragmatic. They've taken consumer-level technologies, 1389 01:30:32,285 --> 01:30:35,909 they've packaged them up in very, um, you know, 1390 01:30:36,454 --> 01:30:39,742 practical, you know, pragmatic ways, and they've built 1391 01:30:40,242 --> 01:30:43,606 some things that are, you know, blasting out US bases in the Middle East as 1392 01:30:44,106 --> 01:30:46,445 we're speaking, right? So you realize that and you're like, fuck, 1393 01:30:47,215 --> 01:30:51,499 okay, we have to change our game. We're not, we're not dead. Right? We still, 1394 01:30:51,628 --> 01:30:54,976 we still got a lot of advantages, but the way the path we're going, 1395 01:30:55,073 --> 01:30:58,196 which is like building interceptors that cost, you know, 100 times more than the thing 1396 01:30:58,228 --> 01:31:01,377 they're intercepting, you know, we're cooked. 1397 01:31:01,603 --> 01:31:04,286 We, we can't go that way, right? And you're like, fuck, okay, we got to 1398 01:31:04,786 --> 01:31:08,213 do something. We got to do something that gets us ahead in this thing that 1399 01:31:08,713 --> 01:31:11,676 we've been asleep on because we've been building these boutique weapons that we've been firing 1400 01:31:11,725 --> 01:31:14,979 at goat herders. And all of a sudden, you know, the opponents are, 1401 01:31:15,740 --> 01:31:19,738 uh, have not been been, you know, sleeping for 100 years. Who would 1402 01:31:20,238 --> 01:31:25,446 you, you know, seeing the roster of sort of US technology 1403 01:31:25,478 --> 01:31:28,776 experts and whatever, people who understand mass production, 1404 01:31:28,793 --> 01:31:31,105 consumer technology, engineering, building factories, 1405 01:31:31,767 --> 01:31:35,308 uh, what is cutting-edge technology right now, where is technology going, 1406 01:31:35,340 --> 01:31:38,696 what's happening with AI, what's happening with, you know, uh, electric motors, 1407 01:31:38,713 --> 01:31:41,714 what's that, you know, you understand the whole range, like, 1408 01:31:41,859 --> 01:31:45,449 who would you put there? Who would 1409 01:31:45,497 --> 01:31:49,536 be your best bet? I understand that people are upset it's Elon, but like— I 1410 01:31:50,036 --> 01:31:51,989 don't— I mean, we don't have an answer. I just feel like we're reaching a 1411 01:31:52,489 --> 01:31:55,371 point with AI and some of this stuff that we just don't have an answer. 1412 01:31:55,436 --> 01:31:59,491 And so that's the problem. And Elon can be in place of 1413 01:31:59,603 --> 01:32:03,210 who we criticized for taking the first steps in that in a large way. 1414 01:32:03,274 --> 01:32:06,384 But what's for me, the reason I mentioned Project Meridian, 1415 01:32:06,432 --> 01:32:11,113 it's because I feel like Elon has unclear status 1416 01:32:11,613 --> 01:32:14,333 this of whether or not he's a special government employee or not, which is covered 1417 01:32:14,833 --> 01:32:18,869 by criminal conflict of interest rules. And so when you're in an aid position, 1418 01:32:18,997 --> 01:32:23,035 you're basically influencing long-term budgets and force structure without 1419 01:32:23,163 --> 01:32:26,512 oversight. And so I see it differently than 1420 01:32:27,012 --> 01:32:30,182 just saying we're ushering in this new technology. This is essentially 1421 01:32:30,214 --> 01:32:33,772 the new Cold War. This is the new space race, and we're gonna bring in 1422 01:32:33,964 --> 01:32:36,740 all of tech and have sit in front of the White House and we can 1423 01:32:36,788 --> 01:32:40,979 meme about it all day. But he has a lack of military experience 1424 01:32:41,525 --> 01:32:45,539 amongst the co-leads combined, you have the other co-leads of this 1425 01:32:45,925 --> 01:32:49,200 alongside commercial stakes. And so there is the perception 1426 01:32:49,216 --> 01:32:52,444 of an oligarchy or self-dealing. I mean, obviously we can 1427 01:32:52,944 --> 01:32:56,313 sit here and laugh at everyone calling him a Nazi and criticize that. But 2 1428 01:32:56,813 --> 01:32:59,589 things are always true at once, I feel like in this timeline, is you have 1429 01:33:00,089 --> 01:33:03,624 the Democrats just unfairly attacking Elon or anybody, 1430 01:33:03,980 --> 01:33:07,508 anybody that's a political enemy. But also there are still things happening 1431 01:33:07,864 --> 01:33:10,195 that are controversial and are not clear-cut. 1432 01:33:11,482 --> 01:33:15,972 And this involves a lot of money and the future of maybe 1433 01:33:16,279 --> 01:33:18,136 like national security. 1434 01:33:20,320 --> 01:33:23,854 Yeah, look, I will tell you that of all the people that are, 1435 01:33:24,554 --> 01:33:28,214 their whole thing is to like take money from the government. I think 1436 01:33:28,714 --> 01:33:32,856 if Elon could avoid being involved here, he would. But as I was explaining 1437 01:33:32,872 --> 01:33:36,823 before, he is so big right now in terms of, you know, his various ventures 1438 01:33:37,323 --> 01:33:40,196 and, and all the things that they're getting involved in that I don't think he 1439 01:33:40,212 --> 01:33:43,408 has a choice. I don't think he, he can choose to stay out 1440 01:33:43,472 --> 01:33:47,118 of these matters because he keeps getting entangled with them. 1441 01:33:47,327 --> 01:33:51,519 And I, I don't think special government employee is the right 1442 01:33:52,019 --> 01:33:55,366 framework here because what can he advise on that he isn't somehow involved in? 1443 01:33:56,712 --> 01:34:00,429 And at the same time, if you don't want the, 1444 01:34:01,198 --> 01:34:04,226

uh, advice of the people that are involved in the thing, 1445 01:34:05,460 --> 01:34:09,209 whose advice are you gonna get, right? And when you say he hasn't fought 1446 01:34:09,709 --> 01:34:12,815 a war, I mean, that's correct, um, but how is 1447 01:34:13,315 --> 01:34:16,019 the advice of the people who have fought wars going, right? Like, they are the 1448 01:34:16,519 --> 01:34:19,592 ones that have been advising to make these Patriot Pack 1, Pack 2, Pack 3 1449 01:34:19,897 --> 01:34:23,361 that cost You know, a multiple of the thing that they're shooting 1450 01:34:23,409 --> 01:34:27,340 down right now, and and we're out, by the way. Right? We cannot manufacture 1451 01:34:27,840 --> 01:34:31,672 them. So what does it matter if you've got the perfect weapon if you 1452 01:34:32,172 --> 01:34:36,373 can't make it? If you're out, right? Like that's what Elon is 1453 01:34:37,031 --> 01:34:40,416 like the best at in the United States by like a 1454 01:34:40,657 --> 01:34:44,299 very wide margin is how to make a complex technology product at 1455 01:34:44,799 --> 01:34:48,772 volume. volume. That, that's like 1456 01:34:49,076 --> 01:34:52,297 the, the thing he's done at Tesla. If you, if you have followed the trajectory 1457 01:34:52,797 --> 01:34:55,775 from the first Roadster to the CyberCab, um, 1458 01:34:56,912 --> 01:35:00,214 he has streamlined the essence of what, you know, a, 1459 01:35:00,438 --> 01:35:04,140 a, a, a car, a consumer vehicle is, to the 1460 01:35:04,640 --> 01:35:08,515 point where, you know, he has probably reduced the cost by a factor of 10 1461 01:35:08,931 --> 01:35:12,506 or, uh, or more. Plus um, 1462 01:35:12,778 --> 01:35:16,130 increase the ability to manufacture it by, I don't know, hundreds, thousands of times. 1463 01:35:16,964 --> 01:35:21,118 That's what you need. And again, like, I, I wish— like, 1464 01:35:21,199 --> 01:35:23,925 I, I have tweeted this. This is not something I'm saying now because it's a 1465 01:35:24,425 --> 01:35:28,159 convenient argument. It is a disaster for the United States that is depending on one 1466 01:35:28,175 --> 01:35:31,624 man. It's terrible, right? 1467 01:35:31,784 --> 01:35:35,698 It is a real problem. It's a huge risk. They say it's 1468 01:35:35,730 --> 01:35:38,504 the bus factor. If you get hit by a bus, right, right, fails. 1469 01:35:38,617 --> 01:35:41,818 That's not the position you want to be in. I wish we had 1470 01:35:42,318 --> 01:35:45,551 more of him, but— We definitely need to chop him in for expertise. 1471 01:35:45,599 --> 01:35:49,687 He has a lot of information in space. Everything is going in that direction of 1472 01:35:50,187 --> 01:35:53,345 warfare. So yeah, and again, 1473 01:35:54,135 --> 01:35:57,245 this is another tweet I've made that whether he likes it or not, he is 1474 01:35:57,342 --> 01:36:02,479 gonna be turned into a weapons dealer. Because when 1475 01:36:06,279 --> 01:36:09,494 the weapons of the future include mass-produced AI 1476 01:36:09,624 --> 01:36:12,940 electric weapons, electric, uh, systems. Um, 1477 01:36:13,376 --> 01:36:16,997 I mean, you know, it's not even a question. 1478 01:36:17,110 --> 01:36:19,987 Like, for all we know, they, they will use the, 1479 01:36:20,003 --> 01:36:23,670 um, Defense Production Act to force him to do it 1480 01:36:23,783 --> 01:36:27,057 if he doesn't— if he doesn't agree to do it by himself. Um, 1481 01:36:27,493 --> 01:36:31,268 anyway, it's, it's, it's a, it's a, it's a hard situation. All I'm saying 1482 01:36:31,285 --> 01:36:34,415 is, if the US does want more Elon Musks, if it, 1483 01:36:34,592 --> 01:36:37,593 if it has not given up on itself and is just willing to just lay 1484 01:36:38,093 --> 01:36:41,178 down and, uh, accept whatever terms are, are, are, um, 1485 01:36:42,228 --> 01:36:45,666 dictated to it. Um, the, 1486 01:36:45,747 --> 01:36:50,008 the path does not involve sending Elon Musk to The Hague because 1487 01:36:50,121 --> 01:36:52,445 you're not just going to get rid of Elon Musk, you're going to get rid 1488 01:36:52,945 --> 01:36:56,351 of anyone who might become Elon Musk, right? You, you, you are 1489 01:36:56,481 --> 01:36:57,973 literally penalizing the entire class. 1490 01:36:59,643 --> 01:37:02,901 Um, and again, I, I don't know how that— you know, you, you tell me 1491 01:37:03,401 --> 01:37:03,615 how you go from there. 1492 01:37:06,560 --> 01:37:10,690 There are no answers here, 1493 01:37:11,190 --> 01:37:14,450 but this topic is why I liked Andrew Yang, because when I first met him, 1494 01:37:14,950 --> 01:37:17,750 he was in Georgetown talking about the revolving door of consultants, why there are no 1495 01:37:18,250 --> 01:37:22,090 in-house tech talent on the government side, and we're not hiring— like, literally all 1496 01:37:22,590 --> 01:37:26,830 tech is outsourced. And so when it becomes more consequential to the timeline 1497 01:37:27,330 --> 01:37:30,370 we're talking in now, it becomes a lot more complicated and scary. And so, 1498 01:37:30,870 --> 01:37:31,851 yeah, I'm excited to see what he has to say. So this is just— but 1499 01:37:32,351 --> 01:37:34,725 you're right, he is a single point of failure as well. But we're not having 1500 01:37:34,758 --> 01:37:38,563 a bunch of Elon Musks certainly anywhere in the world, but mostly not 1501 01:37:38,676 --> 01:37:42,048 inside the walls of the government. You can't really get a lot done with 1502 01:37:42,548 --> 01:37:46,207 the security measures and the bureaucracy, which has presented a lot of problems with Doge 1503 01:37:46,545 --> 01:37:50,190 and some of his other interactions. But the entire system is gonna be 1504 01:37:50,690 --> 01:37:53,996 flawed when tech is more powerful than government and it has to aid everything. 1505 01:37:54,108 --> 01:37:57,561 We don't have a system to support that the way government contracting 1506 01:37:58,061 --> 01:38:01,179 goes. Yep. Well, Alex, 1507

01:38:01,453 --> 01:38:05,461 we are going to be moving forward here. Um, I, I have, 1508 01:38:05,799 --> 01:38:08,971 I have labeled this section, uh, Hack the Planet. 1509 01:38:09,019 --> 01:38:11,704 I appreciate, by the way, just, I want to be clear, I appreciate KK for, 1510 01:38:11,913 --> 01:38:15,190 um, uh, bringing the other side to this. I don't want to— Absolutely. 1511 01:38:15,287 --> 01:38:18,725 We always, we always enjoy all of our point of view, all of our speakers 1512 01:38:18,982 --> 01:38:21,628 with the, with what they bring to the table and the conversations that they bring 1513 01:38:21,660 --> 01:38:24,809 to the table. This, this next section, Alex, that I've labeled here, 1514 01:38:25,004 --> 01:38:29,413 I'm I'm calling Hack the Planet. Hack the planet! 1515 01:38:30,151 --> 01:38:33,522 Great. And, uh, we are going to begin with, 1516 01:38:33,682 --> 01:38:37,695 uh, with an homage to what, what was my favorite 1517 01:38:38,195 --> 01:38:41,322 story of the week, still my favorite story of the week, and was started as 1518 01:38:41,402 --> 01:38:44,757 early as Tuesday. If you do not give me the 1519 01:38:45,257 --> 01:38:49,203 combination to the air shield, Dr. Slotkin will give your daughter back 1520 01:38:49,668 --> 01:38:50,358 her old nose! 1521 01:38:54,380 --> 01:38:57,220 No! Where did you get that? 1522 01:38:58,350 --> 01:39:01,482 All right, I'll tell, I'll tell. No, Daddy, 1523 01:39:01,514 --> 01:39:05,005 no, you mustn't. You're right, my dear. I'll miss your 1524 01:39:05,505 --> 01:39:08,406 new nose, but I will not tell him the combination no matter what. 1525 01:39:09,400 --> 01:39:12,932 Very well. Dr. Slotkin, do your worst. 1526 01:39:13,303 --> 01:39:14,238 My pleasure. 1527 01:39:18,154 --> 01:39:21,953 No! Wait, wait! I'll tell, 1528 01:39:22,485 --> 01:39:25,640 I'll tell. I knew it would work. 1529 01:39:26,563 --> 01:39:28,110 All right, give it to me. 1530 01:39:29,790 --> 01:39:34,108 The combination is— 1. 1531 01:39:34,674 --> 01:39:38,422 1. 1. 2. 2. 3. 1532 01:39:38,797 --> 01:39:41,770 3. 3. 4. 4. 4. 1533 01:39:44,285 --> 01:39:47,595 5. 5! 5! So the combination is 1534 01:39:48,182 --> 01:39:49,682 1, 2, 3, 4, 1535 01:39:51,589 --> 01:39:55,222 5. That's the stupidest combination I ever heard in my life! That's the kind of 1536 01:39:55,722 --> 01:39:58,178 thing an idiot would have on his luggage! Thank you, Your Highness. 1537 01:39:58,388 --> 01:40:01,776 Well, did it work? Where's the key? 1538 01:40:02,034 --> 01:40:05,477 It worked, sir! We have the combination! Great! Now we can take every 1539 01:40:05,977 --> 01:40:09,239 last breath of fresh air from Planet Druidia! What's the combination? 1540 01:40:09,563 --> 01:40:12,888 1, 2, 3, 4, 5. 1, 2, 3, 1541 01:40:12,905 --> 01:40:16,324 4, 5? Yes. That's amazing! I've got the same combination on my luggage! 1542 01:40:20,330 --> 01:40:23,399 The internet goes on high alert after discovery of 8 1543 01:40:23,415 --> 01:40:25,620 compromised TLS certs. 1544 01:40:26,632 --> 01:40:30,697 Security personnel around the world are on high alert Tuesday following 1545 01:40:30,729 --> 01:40:33,995 the discovery of 8 TLS certs with private keys 1546 01:40:34,028 --> 01:40:37,692 so weak they could allow anyone to mint trusted browser and app credentials 1547 01:40:38,192 --> 01:40:41,559 for any site on the internet. As late as Tuesday 1548 01:40:42,059 --> 01:40:45,925 morning, browser makers and some major apps and programming languages had blocked all 1549 01:40:46,198 --> 01:40:49,248 8 of the intermediate certificates. There are no reports 1550 01:40:49,748 --> 01:40:53,502 of certificate transparency logs, which track the issuance of all browser-trusted certificates, 1551 01:40:53,823 --> 01:40:57,258 showing the compromised keys being used maliciously. The threat remains 1552 01:40:57,290 --> 01:41:00,661 far from over, however, because it will be hard and time-consuming to 1553 01:41:01,161 --> 01:41:04,482 update other key parts of the internet, such as Linux distributions 1554 01:41:04,835 --> 01:41:06,729 and embedded and Internet of Things devices. 1555 01:41:07,948 --> 01:41:11,815 The eight intermediate certificates belong to an India-based certificate 1556 01:41:12,315 --> 01:41:16,226 authority, eMudra Limited. The informal advisory published 1557 01:41:16,726 --> 01:41:20,718 to eMudraWeHaveAProblem.com over Tuesday posted the message 1558 01:41:20,798 --> 01:41:24,407 "We have a problem here, Wayne." The words were digitally signed 1559 01:41:24,907 --> 01:41:28,354 using the private key of one of eMudra's certificates. With proof 1560 01:41:28,402 --> 01:41:31,963 that the reporting party had control of the key, defenders scrambled because 1561 01:41:32,011 --> 01:41:35,334 the same key could be used to sign. unauthorized certificates that 1562 01:41:35,834 --> 01:41:39,606 would be trusted by millions of servers worldwide. Now, for those who aren't understanding 1563 01:41:39,638 --> 01:41:43,011 this story, I will explain— Alex and I will explain the 1564 01:41:43,511 --> 01:41:46,914 consequences in easier terms here, but this is my 1565 01:41:47,042 --> 01:41:50,158 favorite story and has been everyone's favorite story of a 1566 01:41:50,334 --> 01:41:53,611 particular class of citizen. This will be 1567 01:41:53,627 --> 01:41:57,160 a massive cleanup job, bigger than Heartbleed. Filippo Valsodora, 1568 01:41:57,305 --> 01:42:02,091 a cryptography engineer and cryptography maintainer for the open-source 1569 01:42:02,591 --> 01:42:04,732 Go programming language told ours, "Quote there'll be very very long a very very long 1570 01:42:04,844 --> 01:42:08,756 tail of IoT devices, Docker containers, forgotten Linux machines, 1571 01:42:08,868 --> 01:42:12,330 unmaintained CLI tools that ship their own copy of the roots. 1572 01:42:13,100 --> 01:42:16,194 Also, Linux distributions are still rolling out updates. Some of them are 1573 01:42:16,694

--> 01:42:19,528 not ready yet, and then system administrators will need to apply them to a lot 1574 01:42:20,028 --> 01:42:24,241 of machines." Heartbleed is the name of a critical vulnerability in the OpenSSL 1575 01:42:24,741 --> 01:42:27,992 cryptography library. In 2014, it came under mass exploitation. 1576 01:42:28,344 --> 01:42:31,999 Exploits let many unaffiliated hackers decrypt TLS protected. 1577 01:42:32,080 --> 01:42:36,076 affected exchanges between parties. They also recovered private TLS certificate keys, 1578 01:42:36,285 --> 01:42:39,751 session cookies, and passwords. The attacks compromised thousands of servers, 1579 01:42:39,767 --> 01:42:42,672 some belonging to large companies and government agencies around the world. 1580 01:42:43,010 --> 01:42:46,605 Recovery required a long series of steps that took years, steps including 1581 01:42:46,701 --> 01:42:50,649 patching OpenSSL on all affected devices, reissuing millions 1582 01:42:51,149 --> 01:42:54,517 of TLS keys, revoking old certificates, rotating passwords and session tokens, 1583 01:42:54,566 --> 01:42:58,545 and updating embedded devices and other forgotten systems. Something that was extremely time 1584 01:42:58,737 --> 01:43:00,868 and resource intensive, especially at internet scale. 1585 01:43:01,702 --> 01:43:04,715 Every time you need a lot of different people in different contexts to do something, 1586 01:43:04,843 --> 01:43:08,449 it will take a long time, Valsadora explained. This time it's worse because 1587 01:43:08,949 --> 01:43:12,984 there's more context, more people, and more heterogeneity. The compromised 1588 01:43:13,176 --> 01:43:17,151 eMudra keys were defective at a level that's challenging the imagination 1589 01:43:17,567 --> 01:43:20,788 of many of the engineers studying the incident. 1590 01:43:21,333 --> 01:43:25,564 A normal size for a NIST P384 elliptic curve cryptography 1591 01:43:25,596 --> 01:43:28,849 private key contains roughly 384 bits of entropy. 1592 01:43:29,749 --> 01:43:33,475 A size allowing almost 2,384, or a number 1593 01:43:33,975 --> 01:43:37,072 with 116 digits of possible different, uh, 1594 01:43:37,458 --> 01:43:41,023 combinations. The private key sizes of the compromised 1595 01:43:41,072 --> 01:43:44,589 keys ranged from 3 to 8 bits. Astonishingly, 1596 01:43:45,103 --> 01:43:48,074 the private keys were 4, 56, 1597 01:43:48,556 --> 01:43:50,853 66, 115, 1598 01:43:51,592 --> 01:43:55,189 155, 156, 230, 1599 01:43:55,928 --> 01:43:59,412 and 237. Now I want to explain this sentence, just so 1600 01:43:59,460 --> 01:44:02,961 you know. That is not a string of numbers that you have 1601 01:44:03,073 --> 01:44:07,296 to all get like a lottery. Any single one of those numbers— 4 1602 01:44:08,147 --> 01:44:11,294 wins the lottery, 56 wins the lottery— and I will 1603 01:44:11,794 --> 01:44:14,762 explain, uh, the significance of what this means. Um, 1604 01:44:14,971 --> 01:44:18,134 that is ex— that is extremely in

the this should not 1605 01:44:18,182 --> 01:44:21,793 be possible category. Veldora said Maybe some horribly 1606 01:44:22,293 --> 01:44:25,785 dangerous key generation interface has a size parameter and they set it to 1 1607 01:44:25,961 --> 01:44:29,247 by mistake. Ben Cartwright-Cox, director of PGP 1608 01:44:29,747 --> 01:44:33,207 Tools, said, my guess is that it's example code that accidentally leaked into production, 1609 01:44:33,415 --> 01:44:36,541 but hard to know. It's bonkers that no one noticed. 1610 01:44:37,503 --> 01:44:41,078 At this early stage, the most immediate threat appears to be for those 1611 01:44:41,110 --> 01:44:44,877 using VPNs, especially in countries with governments likely to maliciously 1612 01:44:45,377 --> 01:44:49,413 intercept traffic. These are likely many other cases— there are likely many 1613 01:44:49,445 --> 01:44:52,635 other cases for concern. End users largely have no recourse other than 1614 01:44:53,135 --> 01:44:56,728 to Hold off using many non-browser forms of TLS communication until they 1615 01:44:57,228 --> 01:45:00,644 receive assurances updates are in place. As noted, most browsers have 1616 01:45:00,740 --> 01:45:04,014 already blocked the intermediate certificates. The exception appears to be Edge 1617 01:45:04,191 --> 01:45:07,658 running on Windows 10. That's Microsoft Alex telling you, "Ha ha! 1618 01:45:07,802 --> 01:45:11,189 We got you now," which at the time you 1619 01:45:11,689 --> 01:45:14,960 better upgrade, which at the time this article went live was still connected to 12 1620 01:45:15,460 --> 01:45:19,165 vulnerable hosts. In response to questions asking why and if the company. Has plans 1621 01:45:19,665 --> 01:45:22,724 to do so. A Microsoft representative for the company will be in touch, did not 1622 01:45:23,224 --> 01:45:26,588 have anything to share. Meanwhile, Signal, WhatsApp, WireGuard, and other apps 1623 01:45:27,088 --> 01:45:30,517 that don't rely on TLS for confidentiality aren't affected. This page 1624 01:45:30,597 --> 01:45:33,644 and the emudrawehaveaproblem.com site show apps 1625 01:45:34,144 --> 01:45:37,428 known to be patched so far. The incident is the latest to expose a fatal 1626 01:45:37,928 --> 01:45:41,019 flaw in the TLS regimen. It is as strong as its weakest link, 1627 01:45:41,083 --> 01:45:44,082 which in which case is emudra. Fortunately, 1628 01:45:44,210 --> 01:45:47,963 transparency logs account for much of that failure, but this measure requires 1629 01:45:47,995 --> 01:45:51,750 parties to actively monitor massive feeds and work— and works only after 1630 01:45:51,830 --> 01:45:55,778 unauthorized certificates have been issued. eMudra representatives weren't immediately 1631 01:45:56,278 --> 01:45:59,742 available to answer questions. One caveat: exploiting compromised certificates 1632 01:45:59,774 --> 01:46:03,096 requires an active adversary-in-the-middle attack, meaning the attacker must act 1633 01:46:03,596 --> 01:46:06,995 as

a proxy between 2 TLS parties and receive encrypted traffic, decrypt it, 1634 01:46:07,075 --> 01:46:11,248 re-encrypt it, and send it to the other party. Such capabilities in some cases are 1635 01:46:11,748 --> 01:46:15,292 usually limited to the most well-resourced attackers— people with control of a local 1636 01:46:15,792 --> 01:46:19,358 network at a coffee shop, university, or other organization. face much less of a challenge. 1637 01:46:19,406 --> 01:46:22,758 So to explain that story to those who 1638 01:46:22,806 --> 01:46:24,771 are not already up to speed, 1639 01:46:26,892 --> 01:46:30,105 the internet has a security layer. 1640 01:46:30,250 --> 01:46:33,688 It has a responsibility. It has a difficult task of 1641 01:46:33,704 --> 01:46:38,010 trying to ensure trust in a world where nobody 1642 01:46:38,139 --> 01:46:41,591 really has any idea what's going on. I have a bank. 1643 01:46:41,688 --> 01:46:44,760 You all have a bank. How do you know that when you go to your 1644 01:46:44,895 --> 01:46:48,159 bank, you are actually talking to your bank and not 1645 01:46:48,223 --> 01:46:52,167 some malicious actor. We all know of the 1646 01:46:52,199 --> 01:46:55,517 phishing scams, the types where you get an 1647 01:46:55,645 --> 01:46:59,316 email that pretends to be PayPal, takes you to a PayPal-looking 1648 01:46:59,364 --> 01:47:02,330 site, and asks you to fill out your username and password. 1649 01:47:02,394 --> 01:47:06,530 And those are things where we understand if we're just a little bit more perceptive, 1650 01:47:07,283 --> 01:47:10,666 we can get around worrying about these problems. But when you 1651 01:47:11,291 --> 01:47:14,785 go to chase.com for Chase Bank, or you go to 1652 01:47:14,801 --> 01:47:18,586 some other website where you are going to enter any sort of important 1653 01:47:18,907 --> 01:47:23,288 personally identifiable or financial information, you personally 1654 01:47:23,304 --> 01:47:27,075 have a very limited ability to know that 1655 01:47:27,575 --> 01:47:31,167 the computer you are talking to on the other end is legitimate. 1656 01:47:31,745 --> 01:47:36,430 And one of the ways that we have attempted to handle 1657 01:47:36,543 --> 01:47:40,555 this problem is by having a trust network. 1658 01:47:41,277 --> 01:47:45,385 And so there is a set of companies Who 1659 01:47:45,885 --> 01:47:49,512 are given this trust to issue what they call 1660 01:47:49,737 --> 01:47:53,767 certificates. And these certificates can be purchased by 1661 01:47:53,895 --> 01:47:56,721 people. So if you set up a website, 1662 01:47:57,123 --> 01:48:01,281 if you, you know, set up myownwebsite.com and 1663 01:48:01,781 --> 01:48:05,520 you wanna make sure that people connecting to it can verify that 1664 01:48:06,020 --> 01:48:09,807 it is, you know, you and your official company at mywebsite.com, 1665 01:48:10,737 --> 01:48:15,031 you can actually purchase a certificate from one of these companies that 1666 01:48:15,159 --> 01:48:18,957 through a couple pieces of configuration will certify 1667 01:48:19,726 --> 01:48:22,787 that the computer that is running the, 1668 01:48:23,091 --> 01:48:27,145 the web server that serves the webpage is 1669 01:48:27,481 --> 01:48:30,975 actually legitimate, that, that the company itself has talked to another 1670 01:48:31,167 --> 01:48:35,061 official company, an authorized trusted company. They've done all 1671 01:48:35,253 --> 01:48:38,572 of the, the certification work themselves. This computer that 1672 01:48:38,620 --> 01:48:42,182 you are doing it, you know, that's running the website is actually 1673 01:48:42,682 --> 01:48:46,130 serving you the page. Here is the proof. And this is all done through mathematics 1674 01:48:46,178 --> 01:48:49,837 and cryptography. But cryptography operates very 1675 01:48:49,885 --> 01:48:52,982 much like a password. So if you know the old joke that, 1676 01:48:53,062 --> 01:48:55,935 you know, password123 is a weak password, 1677 01:48:56,769 --> 01:49:00,348 that is the effect of what we have here, where the trusted company, 1678 01:49:00,765 --> 01:49:04,423 one of the trusted companies able to create these certificates to verify 1679 01:49:04,552 --> 01:49:08,048 that the computer you're talking to is, is a correct computer, 1680 01:49:08,625 --> 01:49:10,437 chose numbers such as 4, 1681 01:49:11,191 --> 01:49:14,543 56, 66, 115, 1682 01:49:14,960 --> 01:49:18,312 155, 156, 230, and 237 1683 01:49:18,969 --> 01:49:21,872 as the password for these certificates. 1684 01:49:22,257 --> 01:49:26,619 And ideally, the password should be exceptionally 1685 01:49:26,667 --> 01:49:30,291 longer. We're talking numbers that, uh, you know, of a large number 1686 01:49:30,791 --> 01:49:34,383 of digits. And some people notice, if they 1687 01:49:34,883 --> 01:49:38,797 have some knowledge, that these aren't prime. Elliptic curve cryptography doesn't require 1688 01:49:39,297 --> 01:49:42,953 prime numbers in this to work. But just know 1689 01:49:42,969 --> 01:49:46,580 that one of the numbers, which is— if you see the number 4 floating around 1690 01:49:46,644 --> 01:49:49,629 on Twitter this last week and you didn't know what 4 was about, 1691 01:49:50,191 --> 01:49:54,187 the number 4 just floating around is probably poking fun at the fact that 1692 01:49:54,203 --> 01:49:57,818 one of these people's keys is 4. And what this means, effectively is that 1693 01:49:58,318 --> 01:50:01,480 any one of you with sufficient capabilities, even using just your, 1694 01:50:01,931 --> 01:50:06,110 your LLM friends now, can create one of these certificates 1695 01:50:06,610 --> 01:50:11,429 that looks like it was authorized by these companies. Now, to use 1696 01:50:11,526 --> 01:50:14,729 this, to exploit it as a malicious actor requires a lot of work. 1697 01:50:16,041 --> 01:50:19,524 It's very difficult to necessarily utilize this for 1698 01:50:20,024 --> 01:50:25,462 a benefit. So if you go to

chase.com, you'll see a certificate 1699 01:50:25,962 --> 01:50:28,654 If you're at your home, there's, you know, almost zero chance, 1700 01:50:28,751 --> 01:50:32,890 nearly zero chance that, that this, this, uh, certificate problem is 1701 01:50:32,906 --> 01:50:36,261 affecting you. But if you go to 1702 01:50:36,778 --> 01:50:40,375 a public Wi-Fi or any— you go to your friend's house 1703 01:50:40,521 --> 01:50:43,760 and you use their Wi-Fi, and let's say your friend is a, uh, 1704 01:50:44,210 --> 01:50:46,993 technically savvy but also terrible human being, 1705 01:50:47,909 --> 01:50:52,124 they could technically send you to a complete 1706 01:50:52,624 --> 01:50:56,485 completely different location than chase.com. They could send you to their own web server and 1707 01:50:56,517 --> 01:51:00,202 they could have a spoofed certificate now. And your 1708 01:51:00,346 --> 01:51:04,304 browser would say, before these updates, hey, everything is okay. 1709 01:51:04,416 --> 01:51:08,486 This is legitimately Chase. And you could then put in your information 1710 01:51:08,582 --> 01:51:12,059 and basically be phished. Other ways that 1711 01:51:12,107 --> 01:51:15,536 they could manipulate it is they could do what's called a man-in-the-middle attack, which is 1712 01:51:15,552 --> 01:51:19,387 where they give you a certificate that looks valid, intercepts your code, 1713 01:51:19,676 --> 01:51:23,162 sends it off to the real chase.com, but they're able to see everything 1714 01:51:23,194 --> 01:51:27,420 you're doing in the middle. But that requires them to have extensive entry 1715 01:51:27,516 --> 01:51:30,119 into your network, which means they kind of have to be living with you, 1716 01:51:30,505 --> 01:51:33,847 or you have to be in like one of those public Wi-Fi places. So in 1717 01:51:33,959 --> 01:51:37,671 general, the average person is not going to be too 1718 01:51:37,719 --> 01:51:40,932 much at risk of dealing with this. But one of the things that they mentioned 1719 01:51:41,432 --> 01:51:45,636 in this article, which is extremely true, is that a lot of devices Internet 1720 01:51:46,136 --> 01:51:49,278 of Things devices, other types of devices that don't get updates that 1721 01:51:49,778 --> 01:51:53,584 often also might have this flaw because one of the ways this trust 1722 01:51:54,084 --> 01:51:57,333 network works is that all of the trusted people, 1723 01:51:57,446 --> 01:52:00,976 all of these companies trusted to provide this security, 1724 01:52:02,920 --> 01:52:06,231 they're updated and sent a list. And if the devices are no 1725 01:52:06,247 --> 01:52:09,965 longer getting updates, that means this company, which now has these trusted 1726 01:52:09,997 --> 01:52:13,417 certificates with numbers like 4 as the password for 1727 01:52:13,529 --> 01:52:16,853 them, well, they have to 1728 01:52:17,353 --> 01:52:20,899 be updated so that they're no longer trusted. And that process requires you to 1729 01:52:20,915 --> 01:52:24,030 identify every such device, then get access to every 1730 01:52:24,062 --> 01:52:27,257 such device. And then if they are, you know, well past 1731 01:52:27,338 --> 01:52:30,469 end of life, you might still be using them. Companies might still be 1732 01:52:30,485 --> 01:52:33,857 using them somewhere. And now you 1733 01:52:34,357 --> 01:52:36,394 can't update them or you have to find them in order to update them or 1734 01:52:36,490 --> 01:52:40,199 updating them is a questionable thing. And you might now have to do the process 1735 01:52:40,699 --> 01:52:43,538 of replacing them Instead of updating them. And this becomes a problem because some of 1736 01:52:43,587 --> 01:52:47,592 these devices are coded 1737 01:52:47,689 --> 01:52:51,186 in a way where they will accept communications from 1738 01:52:51,686 --> 01:52:56,040 certified locations. So you could mimic the certification 1739 01:52:56,136 --> 01:52:58,518 using this hack and then control the device, 1740 01:52:58,890 --> 01:53:02,335 presumably. Again, average people probably don't need to 1741 01:53:02,367 --> 01:53:05,780 worry about that, but this is panicking, of course, the broader 1742 01:53:06,815 --> 01:53:11,034 internet at large because the entire 1743 01:53:11,534 --> 01:53:15,258 trust layer has now had a vulnerability that is all human-induced. Now, Alex, 1744 01:53:15,710 --> 01:53:19,937 my favorite speculation as to what has possibly caused this is 1745 01:53:19,986 --> 01:53:23,833 that they took the P-348 specification as 1746 01:53:23,866 --> 01:53:27,260 to how many bits of entropy they would be provided and thought that 1747 01:53:27,760 --> 01:53:29,898 that was the random number that they would be generating up to. 1748 01:53:32,280 --> 01:53:35,507 And still, I mean, if, if you were 1749 01:53:36,007 --> 01:53:40,866 told to pick random numbers in the range of 0 to 384 or 255 1750 01:53:41,366 --> 01:53:44,593 is the other sum, I assume. Would 4 be 1751 01:53:44,611 --> 01:53:46,929 one of them? Now, 1752 01:53:47,481 --> 01:53:51,213 interestingly enough, interestingly enough, for those who are actually 1753 01:53:51,713 --> 01:53:53,776 very interested, if you happen to be interested in this sort of thing, 1754 01:53:56,098 --> 01:53:59,703 the numbers that you could have chosen are astronomically large. 1755 01:53:59,881 --> 01:54:03,825 You wouldn't be able to even use English language to spell 1756 01:54:03,858 --> 01:54:06,291 them out if you saw them in front of you. You'd have to have some 1757 01:54:06,409 --> 01:54:10,399 computer tell you how to pronounce this using the English language. 1758 01:54:10,464 --> 01:54:13,996 You know, some, some, you know, godforsakenly long number 1759 01:54:14,076 --> 01:54:16,645 with lots of zeros. Lots and lots and lots and lots and lots of digits. 1760 01:54:16,725 --> 01:54:19,952 Yes. Yes. Yes. Now, interestingly enough, 1761 01:54:20,562 --> 01:54:24,126 the funny thing is, is that nobody

attempting to 1762 01:54:24,527 --> 01:54:29,151 attack a system like this would ever imagine that 4 would 1763 01:54:29,651 --> 01:54:35,284 be one of the numbers, which is why this wasn't discovered until 1764 01:54:35,784 --> 01:54:38,268 just recently that nobody even bothered to check a number as small as 4. 1765 01:54:39,439 --> 01:54:43,274 Gator? Yes? Might I suggest? 1766 01:54:43,692 --> 01:54:47,382 Yes! That if a 1767 01:54:47,719 --> 01:54:51,089 state agency, or, I mean, let's face it, like, anyone in 1768 01:54:51,137 --> 01:54:54,331 their basement, but also a state-level agency, 1769 01:54:54,443 --> 01:54:59,674 had discovered this fact and 1770 01:54:59,883 --> 01:55:03,282 kept it to themselves, Mm-hmm. That this would yield a 1771 01:55:03,782 --> 01:55:07,691 number of superpowers. Yes, absolutely. Absolutely. 1772 01:55:07,755 --> 01:55:11,555 That they would be sitting on for years. Yes, 1773 01:55:12,148 --> 01:55:16,540 yes. I'm just referring to the arbitrariness of nobody 1774 01:55:16,653 --> 01:55:20,244 in the public having to, you know, the regular environment. 1775 01:55:20,324 --> 01:55:23,273 Yeah, no, so for example, when I was researching Soul Typhoon, 1776 01:55:24,428 --> 01:55:28,674 one of the puzzles was that HTTPS, 1777 01:55:29,131 --> 01:55:32,041 basically their inability to issue arbitrary certificates, 1778 01:55:34,279 --> 01:55:37,706 was would be a barrier to their ability 1779 01:55:38,206 --> 01:55:40,276 to have complete control over the U.S. network. 1780 01:55:41,004 --> 01:55:46,005 But the the 1781 01:55:46,505 --> 01:55:49,146 U.S. Secret Services, I don't remember the CISA, I believe, 1782 01:55:50,608 --> 01:55:53,714 issued a warning that told everybody to switch 1783 01:55:54,214 --> 01:55:57,260 to Signal basically. Um, implying that they did 1784 01:55:57,760 --> 01:56:00,744 have that power. And, and I just could not figure out, 1785 01:56:01,435 --> 01:56:04,180 you know, how to bridge that gap, 1786 01:56:05,128 --> 01:56:08,002 you know. But, and I'm not saying that they, they had figured this out. 1787 01:56:08,162 --> 01:56:10,105 All I'm saying is that they totally could have. 1788 01:56:10,346 --> 01:56:13,316 There would be nothing to stop them. 1789 01:56:13,718 --> 01:56:17,058 I mean, it is so bad. Um, yeah, so, so effectively, 1790 01:56:17,893 --> 01:56:20,381 I guess, for just to bake it, to boil it down for, for people, 1791 01:56:21,292 --> 01:56:24,599 The security of the internet is based on an 1792 01:56:25,099 --> 01:56:28,228 assumption that nobody would be stupid enough to do what 1793 01:56:28,728 --> 01:56:32,616 these guys did. We're like, 1794 01:56:32,665 --> 01:56:36,280 look, yes, I guess technically somebody could be stupid enough 1795 01:56:36,780 --> 01:56:40,538 to do that, but then they'd have trouble feeding themselves and other things. So we 1796 01:56:40,843 --> 01:56:44,098 are sort of operating on, on the hope that they would never 1797 01:56:44,227 --> 01:56:47,726 reach the point where, where they're both that stupid 1798 01:56:47,855 --> 01:56:49,209 and able to make that decision. 1799 01:56:52,370 --> 01:56:56,300 And yet here we are. I mean, yes. And, and I was, I was, 1800 01:56:56,800 --> 01:56:59,120 I think I, Alex, I was like about 30 minutes ahead of you on this, 1801 01:56:59,620 --> 01:57:02,620 this one. You, you had the same reaction like, oh no, 1802 01:57:03,120 --> 01:57:05,720 Gator, this is, it's only Tuesday. And I'm like, I know. And it's the best 1803 01:57:06,220 --> 01:57:09,680 story this week. It's the greatest story. Yeah. I don't know. 1804 01:57:09,880 --> 01:57:12,720 This does not get beat. This does not get beat. I, I, I don't like 1805 01:57:13,220 --> 01:57:14,960 this year. I, I like, I mean, there, there's gonna be a lot of stupid 1806 01:57:15,460 --> 01:57:18,787 shit. But I mean, how to beat this? 1807 01:57:19,547 --> 01:57:23,366 Well, I mean, listen, Alex. Alex, this is this is the counterargument 1808 01:57:23,819 --> 01:57:27,617 to Eliezer Yudkowsky in this worry about the the the 1809 01:57:28,117 --> 01:57:31,943 super intelligence outsmarting us all. Buddy, we can outsmart 1810 01:57:32,443 --> 01:57:36,169 ourselves no problem. We can do this. There was a there was a 1811 01:57:36,540 --> 01:57:39,734 there was a tweet I loved, which is like you know here's how 1812 01:57:40,106 --> 01:57:43,266 AI labs worry about security. Oh, what if 1813 01:57:43,766 --> 01:57:47,939 the AIs communicate via some unknown protocol over air-gapped 1814 01:57:48,439 --> 01:57:52,290 computers that in reality SSH key 1815 01:57:52,483 --> 01:57:52,997 form? 1816 01:57:53,928 --> 01:57:57,011 Or disgraced, 1817 01:57:57,541 --> 01:58:00,913 disgraced eMudra drops out of TLS business after 1818 01:58:00,929 --> 01:58:05,425 using cert private key of 4. Oh shit. 1819 01:58:05,811 --> 01:58:08,942 Ars Technica. eMudra, the certificate authority that 1820 01:58:08,958 --> 01:58:12,426 issued 8 intermediate TLS credentials so weak that anyone could 1821 01:58:12,474 --> 01:58:16,307 mint certificates Certificates for any website or mail server had a private encryption 1822 01:58:16,419 --> 01:58:19,113 key of just four, and said it's getting out of the business. 1823 01:58:20,492 --> 01:58:24,036 V. Srinivasan, chairman of the India-based 1824 01:58:24,260 --> 01:58:28,124 eMudra Limited, cited increased costs and complexities of compliance 1825 01:58:28,624 --> 01:58:32,550 requirements paired with the business generating how 1826 01:58:32,694 --> 01:58:36,302 big the numbers are that we would have to type in to stay 1827 01:58:36,802 --> 01:58:40,535 in business. Paired with the business generating just one percent of 1828 01:58:41,035 --> 01:58:44,864 its revenue. He also said that the emergence of organizations providing TLS 1829 01:58:45,364 --> 01:58:48,888 certificates free of charge further reduced returns on investment when compared to other

1830 01:58:49,388 --> 01:58:52,976 products and services. The don't sorry products and 1831 01:58:53,476 --> 01:58:56,792 services, and then the next headline was don't let the door hit you on 1832 01:58:56,824 --> 01:59:00,095 the butt. Quote however, the level of social media amplification 1833 01:59:00,595 --> 01:59:03,959 around the quote however, the level of social media amplification around this incident 1834 01:59:04,459 --> 01:59:08,769 has brought scrutiny that is out of proportion to SSL/TLS's 1835 01:59:09,269 --> 01:59:13,000 place in our business. The company continued in a Thursday post, 1836 01:59:14,640 --> 01:59:18,776 quote, in view of this, after detailing— this 1837 01:59:19,276 --> 01:59:21,109 is literal, you had one job. 1838 01:59:22,272 --> 01:59:25,791 Like, in view of this, after detailed deliberations, it's not 1839 01:59:26,291 --> 01:59:29,583 2 jobs, it's one. Our board has concluded that withdrawing from 1840 01:59:30,083 --> 01:59:33,585 global SSL/TLS is the right decision for eMudra and its shareholders. 1841 01:59:34,853 --> 01:59:38,106 The announcement came 2 days after researchers revealed eMudra had issued 8 1842 01:59:38,606 --> 01:59:42,133 certificates with private keys so defective that they were trivial to discover. Anyone who took 1843 01:59:42,633 --> 01:59:45,986 the time, which was measured in milliseconds, could crack them and then use them to 1844 01:59:46,486 --> 01:59:49,519 generate TLS certificates that would be trusted by all major browsers, as well as an 1845 01:59:49,615 --> 01:59:53,597 untold number of email servers, IoT devices, Docker containers, forgotten Linux 1846 01:59:53,629 --> 01:59:56,969 machines, and unmaintained CLI tools that ship their own copy 1847 01:59:57,065 --> 02:00:00,999 of the roots. Whereas NIST P384 elliptic 1848 02:00:01,499 --> 02:00:05,013 curve cryptography private keys used in the type of certificates eMudra issued are supposed to 1849 02:00:05,513 --> 02:00:09,118 contain roughly 384 bits of entropy. The 8 compromised certificates had 1850 02:00:09,618 --> 02:00:12,821 sizes ranging from 2— or sorry, 3 to 8 bits. Stated differently, 1851 02:00:12,949 --> 02:00:17,806 whereas they were supposed to be, uh, where there were supposed to be 2,384, 1852 02:00:18,015 --> 02:00:21,589 a number with, uh, 116-digit different possibilities, 1853 02:00:23,048 --> 02:00:27,392 the, the strongest of the defective keys had a maximum of just 256. 1854 02:00:28,098 --> 02:00:31,560 The private keys were 456. We've read those. Browser makers blocked 1855 02:00:32,060 --> 02:00:35,472 certificates that were issued with the immediate intermediate certificates within a day or two of 1856 02:00:35,972 --> 02:00:39,656 them being discovered. Emudra almost certainly faced the the prospect of its certificate authority status 1857 02:00:40,156 --> 02:00:43,441 being pulled by the CA Browser Forum, an industry-wide group that sets rules 1858 02:00:43,489 --> 02:00:46,696 for admittance into the TLS certificate program. Remember, all of 1859 02:00:47,196 --> 02:00:51,009 this, as I stated, is our best attempt at making 1860 02:00:51,105 --> 02:00:55,339 something we call a trustless internet that is ultimately 1861 02:00:55,371 --> 02:00:58,995 based on trusting several important people. 1862 02:00:59,604 --> 02:01:02,539 That's about it. That's the best you get in this world right now, 1863 02:01:02,731 --> 02:01:06,595 which is that we have a massive worldwide 1864 02:01:07,095 --> 02:01:10,618 internet that runs on trusting certain groups to just do their jobs. 1865 02:01:10,746 --> 02:01:13,295 And by and large, it hasn't been completely corrupted. 1866 02:01:14,593 --> 02:01:18,456 It's relatively trustworthy through intelligence 1867 02:01:18,956 --> 02:01:22,111 and cryptography and people doing verifications and a lot of it being 1868 02:01:22,223 --> 02:01:25,444 transparent and not opaque in its 1869 02:01:25,944 --> 02:01:28,201 ability to be inspected from the outside. 1870 02:01:28,714 --> 02:01:32,337 And the certifications of the mathematical cryptography 1871 02:01:32,465 --> 02:01:37,168 allow us to have assurances that things are difficult without making 1872 02:01:37,217 --> 02:01:40,572 sure that people have too many secrets that make us doubt 1873 02:01:41,072 --> 02:01:44,281 them. It's a very difficult structure to maintain, but we've 1874 02:01:44,425 --> 02:01:47,540 been doing a decent job so far until 1875 02:01:48,615 --> 02:01:52,067 the number 4. On Wednesday, eMudra provided a brief 1876 02:01:52,567 --> 02:01:56,819 explanation for the debacle. The certificates were created under eMudra's 1877 02:01:57,319 --> 02:02:00,792 ECC root in a key generation ceremony held on April 30th, 1878 02:02:00,825 --> 02:02:04,423 2025, using FIPS 140-2 Level 1879 02:02:04,455 --> 02:02:08,505 3 hardware security modules. The company said a flaw in the KGC 1880 02:02:08,554 --> 02:02:13,080 caused a wrap-unwrap step to produce a second set of weak keys alongside 1881 02:02:13,580 --> 02:02:17,526 the original strong keys. The weak keys were the ones published to the browser root 1882 02:02:18,026 --> 02:02:20,251 stores. eMudra is working to establish the root cause. 1883 02:02:22,034 --> 02:02:25,379 So that article came out as 1884 02:02:25,412 --> 02:02:28,628 they said they were leaving because of course this makes them look 1885 02:02:29,128 --> 02:02:32,900 bad. But the The backend conversation, Alex, 1886 02:02:32,964 --> 02:02:36,154 I don't know if you've caught wind of some of this, was also pretty funny. 1887 02:02:36,508 --> 02:02:40,393 Now, we agree, Alex, I think, that the idea 1888 02:02:40,442 --> 02:02:43,984 that this is— the 1889 02:02:44,000 --> 02:02:47,574 response in social media and in the regular media 1890 02:02:48,074 --> 02:02:51,918 was undue focus on something that's only 1% of their business. 1891 02:02:52,321 --> 02:02:56,192 It might be 1% of your business, but the

responsibility was to the entire 1892 02:02:56,692 --> 02:02:59,960 internet and the The amount of damage that you've caused with your 1% business 1893 02:03:00,232 --> 02:03:03,855 is astronomical. I don't know how many times the 1% of their 1894 02:03:04,355 --> 02:03:08,039 business they've done in damage, Alex, but it's pretty hefty. Now, 1895 02:03:08,728 --> 02:03:12,143 the backend conversations, Alex, are funny as well. 1896 02:03:12,528 --> 02:03:16,102 So on the Mozilla bug 1897 02:03:16,279 --> 02:03:18,699 channel, so this is Bugzilla, 1898 02:03:19,885 --> 02:03:22,979 where the browser people all discuss these important things. 1899 02:03:24,110 --> 02:03:27,533 Emudra responded to this, the issue that was posted 1900 02:03:28,033 --> 02:03:32,216 about this certificate problem. And this 1901 02:03:32,716 --> 02:03:36,079 was like 2 days after the event where the browsers were still looking for 1902 02:03:36,160 --> 02:03:39,459 answers. The person from Emudra said, 1903 02:03:39,637 --> 02:03:42,934 there are a lot of questions here, as most of you can imagine. We have 1904 02:03:43,434 --> 02:03:47,687 been subject to intense social media scrutiny and analysis by investors, customers over 1905 02:03:48,187 --> 02:03:51,762 the last few days. And it has made doing any meaningful analysis of the event 1906 02:03:52,262 --> 02:03:55,885 pretty much impossible. The event happened on April 30th, 2025, 1907 02:03:55,950 --> 02:03:58,594 and was done by a set of people who worked in the PKI industry for 1908 02:03:59,094 --> 02:04:02,217 several years and worked with us for a while. There is a lot of digging 1909 02:04:02,330 --> 02:04:05,485 we need to do to understand the circumstances, why, what, 1910 02:04:05,549 --> 02:04:08,985 and how to share a meaningful report that can hopefully benefit the community. 1911 02:04:09,501 --> 02:04:12,454 I do not know why some community members think we won't be getting back on 1912 02:04:12,486 --> 02:04:16,085 this bug. When we signed up for the browser 1913 02:04:16,585 --> 02:04:20,501 program trust, we also signed up for disclosure obligations to the community, which we will 1914 02:04:20,506 --> 02:04:24,522 Uphold. Now, Alex, if you were— had 1915 02:04:25,022 --> 02:04:28,603 the opportunity to respond to such a post, what might your approach 1916 02:04:28,635 --> 02:04:30,001 be? What might your attitude be? 1917 02:04:33,406 --> 02:04:37,126 Alex, sir? Look, any, any, any email about 1918 02:04:37,626 --> 02:04:41,411 this topic that attempts to maintain some sort 1919 02:04:41,911 --> 02:04:45,421 of professional, you know, tone or whatever, you know, um, 1920 02:04:46,024 --> 02:04:49,147 is, is purely inappropriate. Right? Like, I don't 1921 02:04:49,212 --> 02:04:52,450 know what this person is trying to do. Of course, yes, yes, of course, 1922 02:04:52,483 --> 02:04:54,880 yes, we're working on it. What, dude? 1923 02:04:58,439 --> 02:05:02,047 Like, okay, let's say for some godforsaken reason I am tasked 1924 02:05:02,304 --> 02:05:07,367 with writing this email. The first words is, we fucked 1925 02:05:08,062 --> 02:05:11,427 up and we are so sorry. We are doing 1926 02:05:11,605 --> 02:05:14,644 our best. Okay, there is no excuse for this. 1927 02:05:15,628 --> 02:05:19,256 Ergo, I won't offer one. We are going to 1928 02:05:19,756 --> 02:05:22,562 do anything humanly possible to do, you know, 1929 02:05:23,654 --> 02:05:26,478 to make this however right as possible to make it, 1930 02:05:27,345 --> 02:05:30,443 right? Like, this is what you— you're not like, well, of course we're going to 1931 02:05:30,491 --> 02:05:32,979 provide you with a blah blah blah for FortiGate. 1932 02:05:33,380 --> 02:05:35,707 It's like, 1933 02:05:35,788 --> 02:05:39,078 yes, we're so past like, uh, you know, 1934 02:05:39,543 --> 02:05:43,540 uh, please send us an incident report that we've been talking 1935 02:05:44,225 --> 02:05:46,727 Here are the people working for Chrome's response. 1936 02:05:48,340 --> 02:05:51,883 While we understand that many things are happening in parallel right now, 1937 02:05:52,563 --> 02:05:56,172 quote, social media scrutiny and quote analysis by 1938 02:05:56,204 --> 02:05:59,992 investors and customers are not acceptable explanations for 1939 02:06:00,008 --> 02:06:03,727 delayed responses here. That scrutiny is the predictable response of 1940 02:06:03,936 --> 02:06:08,236 eMudra's own failure, i.e., intermediate CA keys that anyone could recover, 1941 02:06:08,481 --> 02:06:11,864 resulting in anyone being able 1942 02:06:12,364 --> 02:06:15,568 to issue certificates for any domain, not an external circumstance that befell it. Offering it 1943 02:06:16,068 --> 02:06:19,969 as an excuse conveys that reputation management is being prioritized over 1944 02:06:19,985 --> 02:06:23,583 the obligations that eMudra still owes to the billions 1945 02:06:23,952 --> 02:06:27,679 of relying parties its mistake exposed. Those obligations 1946 02:06:27,791 --> 02:06:31,727 do not relax because the consequences are uncomfortable. We expect 1947 02:06:31,871 --> 02:06:35,068 eMudra to prioritize its review of this issue and its security 1948 02:06:35,164 --> 02:06:37,249 response. Which is the most polite, 1949 02:06:38,146 --> 02:06:41,545 who gives a fuck? Like, that's, 1950 02:06:41,994 --> 02:06:45,200 that is effectively the polite version. 1951 02:06:46,145 --> 02:06:49,576 I missed the part where it was like, you know, due to social media scrutiny. 1952 02:06:49,624 --> 02:06:53,696 It's like, what do you expect? What exactly 1953 02:06:54,196 --> 02:06:57,863 would be a reasonable response here? Please walk me through 1954 02:06:58,363 --> 02:07:02,160 it. Explain. Now, now I have, 1955 02:07:02,176 --> 02:07:06,183 I have a screenshot here, Alex, of the press release from eMudra 1956 02:07:06,683 --> 02:07:08,955 where they say they are going to get out of the TLS business. 1957 02:07:11,362 -->

02:07:14,940 And let me say that it's quite fun because their header here, 1958 02:07:15,939 --> 02:07:18,985 their logo is eMudra, Trust Delivered. 1959 02:07:19,904 --> 02:07:22,658 That's their slogan. That's their slogan. 1960 02:07:23,594 --> 02:07:26,886 Mr. V. Saravasan, Chairman, eMudra Limited, 1961 02:07:26,902 --> 02:07:30,839 referred to the recent incident relating to its global SSL/TLS certificate 1962 02:07:31,339 --> 02:07:34,106 and indicated that eMudra handled the recent incident diligently, 1963 02:07:34,686 --> 02:07:38,228 revoking the affected CAs within 14 hours of discovery and 1964 02:07:38,728 --> 02:07:42,655 disclosing promptly. There were no end-entity certificates issued out of 1965 02:07:42,720 --> 02:07:46,000 the affected CAs. Its infrastructure was not compromised. 1966 02:07:46,309 --> 02:07:50,912 No data was breached. All physical and logical 1967 02:07:51,284 --> 02:07:55,476 security controls remained intact, which is great for them, I guess, but doesn't 1968 02:07:55,541 --> 02:07:59,225 even begin to broach the problem that's out there. He added 1969 02:07:59,725 --> 02:08:03,142 that the global Global SSL/TLS is a small part of 1970 02:08:03,271 --> 02:08:06,273 eMudra's business, contributing to about 1% of global revenue. 1971 02:08:06,947 --> 02:08:10,062 Due to increasing data sovereignty and compliance requirements, 1972 02:08:10,319 --> 02:08:13,755 complexity and cost associated with this business increasing over the 1973 02:08:13,771 --> 02:08:17,737 years, this paired with the fact that some of the large entities issue the SSL 1974 02:08:17,801 --> 02:08:21,462 and TLS free of cost, the business potential and ROI of this line of 1975 02:08:21,962 --> 02:08:27,210 business also is considerably reduced for eMudra when measured against our cumulative 1976 02:08:27,710 --> 02:08:31,369 However, the level of social media amplification around this incident has brought 1977 02:08:31,869 --> 02:08:35,479 scrutiny that is out of proportion. I love this excuse 1978 02:08:35,913 --> 02:08:39,590 to the SSL/TLS place in our business. Because of this, 1979 02:08:39,686 --> 02:08:43,765 after detailed deliberations, our board has concluded that we are withdrawing 1980 02:08:43,861 --> 02:08:46,992 from the global SSL/TLS and is the 1981 02:08:47,492 --> 02:08:50,428 right decision for eMudra and its shareholders. For any certifying authorities, 1982 02:08:50,524 --> 02:08:54,122 SSL/TLS TLS in the whole business. For eModere, it is one part of a 1983 02:08:54,622 --> 02:08:57,847 much broader digital solutions portfolio. We will put our energy 1984 02:08:57,911 --> 02:09:01,521 where we create the most value for customers, with a sharpened focus 1985 02:09:02,021 --> 02:09:06,490 on core enterprise solutions and trust services other than SSL and TLS. 1986 02:09:10,163 --> 02:09:14,529 Oh, man, Alex! Now, listen, as far as public 1987 02:09:14,931 --> 02:09:19,261 relations go, I'm going to have to say that if

I were a potential investor, 1988 02:09:19,615 --> 02:09:23,392 this would horrify Me if I knew any, if I had any semblance 1989 02:09:23,892 --> 02:09:28,027 of understanding of what this was basically saying. They're basically 1990 02:09:28,059 --> 02:09:31,154 going, yes, we made an oopsie, 1991 02:09:31,651 --> 02:09:34,602 but we didn't care about this aspect of the business. 1992 02:09:34,874 --> 02:09:38,819 We paid so little attention to it because it was only 1% that 1993 02:09:38,948 --> 02:09:42,668 we are willing to get rid of it. And the costs that 1994 02:09:42,716 --> 02:09:46,357 everyone else are going to have to incur to fix our mistake are 1995 02:09:46,857 --> 02:09:50,153 not our problem. That is how that reads. I don't have any other 1996 02:09:50,653 --> 02:09:54,086 interpretation. And if I were an investor, there is no trust building in that response. 1997 02:09:56,110 --> 02:09:59,362 Look, I think that there 1998 02:09:59,862 --> 02:10:03,564 are 2 things on, on the basis of this that the global certificate 1999 02:10:04,064 --> 02:10:08,436 infrastructure can do. They, they, they can't fix the problem that they are relying 2000 02:10:08,533 --> 02:10:11,925 on a few trusted parties, uh, to secure the internet. 2001 02:10:12,737 --> 02:10:16,219 Uh, that is a, a different discussion for a different day about how the, 2002 02:10:16,284 --> 02:10:19,846 uh, the US, um, stole the internet from 2003 02:10:19,894 --> 02:10:23,264 the people that created it, um, who they funded in the first place. So fair 2004 02:10:23,764 --> 02:10:27,725 enough, uh, but that's a different story, how DNS is fucked and how HTTPS 2005 02:10:27,838 --> 02:10:30,405 is fucked. And, you know, this is one small manifestation of that. 2006 02:10:31,031 --> 02:10:34,802 However, here are my top 2 things that they could 2007 02:10:35,302 --> 02:10:38,686 do. Number one, uh, companies who, 2008 02:10:39,135 --> 02:10:42,730 uh, are a, um, what's the, what's the term 2009 02:10:43,230 --> 02:10:47,141 Here for what Imadra was—a certificate trust authority, 2010 02:10:47,975 --> 02:10:50,782 CA. One percent of it. 2011 02:10:51,424 --> 02:10:54,857 Yeah, companies whose business, 2012 02:10:55,033 --> 02:10:58,290 whose activity as a CA is less than one percent of their business, 2013 02:10:58,627 --> 02:11:01,257 do not qualify for providing that business. 2014 02:11:02,653 --> 02:11:07,385 Number two, the numbers from 2015 02:11:07,401 --> 02:11:10,795 zero to oh, I don't know. So, um, 2016 02:11:12,121 --> 02:11:16,245 you know, 65,535 are not 2017 02:11:16,745 --> 02:11:20,207 available for selection, thank you! 2018 02:11:21,574 --> 02:11:25,080 In fact, all powers of 2 are directly not permitted 2019 02:11:25,597 --> 02:11:30,675 to be— Yes. You can 2020 02:11:31,175 --> 02:11:35,303 knock out a few other easies. Just, it's like, you know, like, if I selected 2021 02:11:35,803 --> 02:11:39,821 4 for my password on Yahoo Mail, I would be denied. 2022 02:11:40,959 --> 02:11:43,942 There's been one upside. 2023 02:11:45,048 --> 02:11:48,303 One upside, Alex. One upside to this affair that I'm aware of. 2024 02:11:49,056 --> 02:11:52,520 You ready for it? Ready for the upside? Hit me. 2025 02:11:52,953 --> 02:11:56,961 Hit me. Owners of Sonos audio speakers 2026 02:11:57,073 --> 02:12:00,938 who have had several years of complaints about the software 2027 02:12:01,884 --> 02:12:05,780 that has come paired with Sonos that they can't override are now exploiting 2028 02:12:06,280 --> 02:12:10,306 Exploiting the eMudra certificate issue to gain access to 2029 02:12:10,806 --> 02:12:14,495 their own Sonos equipment and try to update the 2030 02:12:14,543 --> 02:12:16,212 firmware themselves. That's right. 2031 02:12:17,352 --> 02:12:20,144 That's right. It's not just Sonos, actually. 2032 02:12:20,658 --> 02:12:24,606 There's a lot you can do. It's really a gift, a gift from our 2033 02:12:24,654 --> 02:12:28,185 Indian friends. So there you go. There's one upside to it, Alex. 2034 02:12:28,683 --> 02:12:31,812 So, uh, get it while the kitten's good, right? All right. 2035 02:12:32,357 --> 02:12:35,515 Now, Alex, this is continuing the Hack 2036 02:12:35,563 --> 02:12:40,164 the Planet segment that we are currently in. Hack the planet! 2037 02:12:41,254 --> 02:12:44,685 And because this has been quite a chaotic week, this is not the only 2038 02:12:44,717 --> 02:12:48,324 story, although it was the funniest and probably the 2039 02:12:48,824 --> 02:12:52,380 most impactful, although there are contenders. This is 2040 02:12:52,476 --> 02:12:55,266 another human error, Alex, that is also kind of funny. 2041 02:12:55,859 --> 02:12:59,916 Several Linux kernel vulnerabilities accidentally posted to public kernel 2042 02:12:59,965 --> 02:13:03,317 mailing list. Seemingly as an accident, overnight several 2043 02:13:03,817 --> 02:13:07,247 new Linux kernel vulnerability reports were posted to the public Linux kernel mailing 2044 02:13:07,747 --> 02:13:11,209 list rather than the private security mailing list. The vulnerabilities have 2045 02:13:11,241 --> 02:13:14,786 yet to be patched on mainline, but at least in some cases there are patches 2046 02:13:14,850 --> 02:13:17,481 undergoing testing or queued in the next branches. 2047 02:13:18,171 --> 02:13:21,363 CVE-SS scores are as high as 9.0. 2048 02:13:21,748 --> 02:13:25,533 Trend Micro's Zero Day Initiative sent out another, uh, a number of vulnerability 2049 02:13:26,033 --> 02:13:29,667 reports to the public Linux kernel mailing list rather than the typical 2050 02:13:29,683 --> 02:13:34,411 private security mailing list. This seemingly was a mistake, with Linux's second-in-command 2051 02:13:34,911 --> 02:13:38,513 Greg Crowe-Hartman responding to each report asking about their submission to the 2052 02:13:39,013 --> 02:13:42,407 public mailing list. So that, that one, Alex, is kind of funny. 2053 02:13:42,583 --> 02:13:46,093 Uh, I had an

image, but, uh, I guess I deleted it accidentally, 2054 02:13:46,590 --> 02:13:49,474 where the guy is just like, why? Okay, 2055 02:13:49,554 --> 02:13:53,244 you didn't submit a patch and you're possible posting 2056 02:13:53,744 --> 02:13:56,711 this to the public mailing list. So for those who didn't follow that, just so 2057 02:13:57,211 --> 02:13:59,760 I can explain, basically there's, you know, a, 2058 02:14:00,466 --> 02:14:05,104 a gentleman's agreement when it comes to discovering vulnerabilities 2059 02:14:05,136 --> 02:14:08,731 in software, uh, particularly when you are, uh, trying to be 2060 02:14:09,231 --> 02:14:12,727 a member of, you know, good standing nature in the community. If you find 2061 02:14:13,144 --> 02:14:16,579 a problem, you try to notify people who are responsible for 2062 02:14:16,611 --> 02:14:20,156 fixing it with the details of what you found. You do this just 2063 02:14:20,656 --> 02:14:23,830 as a, you know, a friendly gesture. Sometimes there are— they have what are called 2064 02:14:23,894 --> 02:14:27,761 bug bounties where the company, you know, will give you some sort 2065 02:14:28,261 --> 02:14:31,338 of finder's fee for the bug if you give them appropriate detailed information. 2066 02:14:31,771 --> 02:14:35,830 But with Linux, Linux is one of the largest open source projects in 2067 02:14:36,087 --> 02:14:40,162 the world. It may, you know, Linux runs billions upon 2068 02:14:40,258 --> 02:14:43,901 billions of machines across the internet all sorts of 2069 02:14:44,190 --> 02:14:47,225 small devices to full server, you know, large servers, 2070 02:14:47,755 --> 02:14:51,754 etc. And when a vulnerability 2071 02:14:51,770 --> 02:14:55,126 is found, there are people who maintain Linux, 2072 02:14:55,432 --> 02:14:58,836 the core root Linux. That's a small select group. 2073 02:14:59,286 --> 02:15:02,642 And there are mailing lists that are private where you can notify them. 2074 02:15:02,690 --> 02:15:05,902 You don't always have to submit a fix for it, but you have to give 2075 02:15:06,402 --> 02:15:09,612 them some sort of details about what you found so that they can try to 2076 02:15:10,112 --> 02:15:13,241 make sure that they patch it. This helps mostly keep things secure so 2077 02:15:13,257 --> 02:15:17,175 that the people who can fix it fix it before the bad actors hopefully 2078 02:15:17,271 --> 02:15:20,916 find out the problem. That's, that's the idea. But if you blast the public list 2079 02:15:21,365 --> 02:15:24,545 and you don't provide a fix, now the bad actors are 2080 02:15:24,914 --> 02:15:28,286 weeks ahead because pushing a fix to everyone takes a 2081 02:15:28,786 --> 02:15:32,396 lot longer. So, Alex, more, more fun this week with 2082 02:15:32,896 --> 02:15:37,406 human-caused errors as opposed to worrying about the super intelligent 2083 02:15:37,640 --> 02:15:40,620 Any comments on that one, Alex? 2084 02:15:41,120 --> 02:15:44,520 Great, great. No, no, keep going. 2085 02:15:45,020 --> 02:15:48,600 Uh, we haven't hit PHP yet. Uh, no, we have not. 2086 02:15:49,100 --> 02:15:51,700 We have not. Um, you said you sent me a message that you need to 2087 02:15:52,200 --> 02:15:55,800 bail soon. How, uh, do you want me to— No, no, no, 5:30. I'll just 2088 02:15:56,300 --> 02:16:00,080 bail for— I'll just need to be briefly gone for 10 minutes. You're 3 hours 2089 02:16:00,580 --> 02:16:03,060 behind me, right? Yeah, you're 3 hours behind me. That's right. Oh, sorry, sorry. 2090 02:16:03,560 --> 02:16:05,500 I mean, in 1 hour and 10 minutes, I'll need to be gone. Thank you. 2091 02:16:05,645 --> 02:16:08,436 Now we can do it in public. No need for the DM. Who needs DMs 2092 02:16:08,936 --> 02:16:13,505 when you can just talk about it? All right. 2093 02:16:13,826 --> 02:16:17,114 So Quest Diagnostic reports a cyber 2094 02:16:17,614 --> 02:16:21,429 breach affecting certain patient records. Medical laboratory operator 2095 02:16:21,929 --> 02:16:24,749 Quest Diagnostics said it learned of a cyber breach in one of its systems this 2096 02:16:24,798 --> 02:16:28,182 week, likely compromising some records for patients. In a Thursday regular, 2097 02:16:28,407 --> 02:16:32,241 uh, regulatory filing, Quest Diagnostics reported unauthorized Access 2098 02:16:32,577 --> 02:16:36,380 to its Quantum Lab Service Manager, a platform where certain clients 2099 02:16:36,880 --> 02:16:39,749 can place test orders and receive or look up their results. The company said it 2100 02:16:40,249 --> 02:16:43,005 discovered the hack on Wednesday and notified federal law enforcement. 2101 02:16:43,407 --> 02:16:47,161 How many patients and which particular records have been affected is still unknown. 2102 02:16:47,257 --> 02:16:51,268 Quest Diagnostics said it believes certain patient records were accessed 2103 02:16:51,768 --> 02:16:55,054 in the breach, but that the full scope remains under investigation. The New Jersey-based 2104 02:16:55,554 --> 02:16:58,985 company said it's also working with external cybersecurity experts to assist in its 2105 02:16:59,485 --> 02:17:02,952 response. Quest Diagnostics has seen hacks affecting patient records before. 2106 02:17:03,626 --> 02:17:07,320 In 2016, for example, the company said its data, including lab results and contact 2107 02:17:07,820 --> 02:17:11,479 information of about 34,000 people, was exposed in another online breach. 2108 02:17:12,009 --> 02:17:15,815 In 2019, other personal information belonging to about 11.9 2109 02:17:15,911 --> 02:17:19,733 million Quest patients was affected after a separate hack at American Medical 2110 02:17:19,782 --> 02:17:23,475 Collection Agency, a billing collection service provider. Cyberattacks are 2111 02:17:23,975 --> 02:17:27,630 on the rise across industries beyond health healthcare. Other breaches have recently disrupted anything from 2112 02:17:28,130 --> 02:17:31,575 supermarket chains to core educational applications. Governments and 2113

02:17:32,075 --> 02:17:35,280 companies worldwide are working more broadly on online security updates. 2114 02:17:35,408 --> 02:17:38,936 Urgency erupted this week after an AI-assisted discovery of vulnerabilities in 2115 02:17:39,436 --> 02:17:42,962 a programming language called PHP, which is vital to about 70% of the websites 2116 02:17:43,462 --> 02:17:47,003 globally. We are getting there, don't worry. Now this 2117 02:17:47,051 --> 02:17:51,189 next— this story, Alex, how a massive Chinese crime ring hacked Staples 2118 02:17:51,221 --> 02:17:54,637 to steal millions. This is from 404 Media. A Chinese 2119 02:17:54,669 --> 02:17:57,318 organized crime group hacked into retail giant Staples, 2120 02:17:57,640 --> 02:18:00,867 used that access to fraudulently issue millions of dollars worth 2121 02:18:01,367 --> 02:18:04,785 of gift cards for stores like Target and Sephora. That went on a nationwide 2122 02:18:05,285 --> 02:18:08,927 shopping spree, shipping many of stolen electronics and other products back to China as 2123 02:18:08,959 --> 02:18:12,556 part of a massive retail theft operation, 404 Media has learned. 2124 02:18:13,214 --> 02:18:16,329 The finding, based on a review of court records, reveals the scale at which some 2125 02:18:16,829 --> 02:18:20,215 of organized crime groups are fleecing U.S. retailers, with entire warehouses 2126 02:18:20,263 --> 02:18:23,808 dedicated to house causing the merchandise. The hack of Staples, which has not 2127 02:18:24,308 --> 02:18:27,466 been previously reported, signals just how sophisticated some of these operations 2128 02:18:27,802 --> 02:18:31,636 have become. In October 2023, Homeland Security 2129 02:18:31,684 --> 02:18:35,326 Investigations received information from an office supply company saying that 2130 02:18:35,342 --> 02:18:38,935 hackers had targeted the company and issued more than 15,000 gift cards 2131 02:18:38,967 --> 02:18:42,817 worth over $5 million, according to a search warrant application unsealed last 2132 02:18:43,317 --> 02:18:46,747 week. The document does not name the company, but 404 Media determined it 2133 02:18:47,247 --> 02:18:50,900 was Staples based on various clues. The document says the victim company is headquartered 2134 02:18:51,400 --> 02:18:54,762 in Massachusetts, uses a company called BlackHop Network to fulfill its gift card 2135 02:18:55,262 --> 02:18:58,384 orders, and the hacked company has an internal customer support system called 2136 02:18:58,432 --> 02:19:02,262 Nexus. All of those identifying features match Staples, and 404 2137 02:19:02,294 --> 02:19:06,188 Media found a reference to Nexus on a former Staples executive's LinkedIn profile. 2138 02:19:07,021 --> 02:19:10,675 The document says Staples believed Nexus was only accessible to the company's internal 2139 02:19:10,707 --> 02:19:15,322 computers, but that the hackers found a way to access it from 2140 02:19:15,822 --> 02:19:18,489 the outside. The accessors performed multiple heap dump requests, which is a way to capture 2141 02:19:18,553 --> 02:19:21,861 all of the data held in an app's memory at a particular point in time, 2142 02:19:21,877 --> 02:19:25,121 according to the document. Using the returned information, the hackers 2143 02:19:25,137 --> 02:19:28,733 were able to then hijack the sessions of logged-in Staples customer service representatives 2144 02:19:29,167 --> 02:19:32,683 to then issue a massive wave of gift cards, it adds. The hackers sent 2145 02:19:33,183 --> 02:19:36,296 the gift cards to various email accounts they controlled. After this, members of 2146 02:19:36,796 --> 02:19:39,138 the gang went and spent many of the gift cards on electronics at stores stores 2147 02:19:39,638 --> 02:19:42,921 like Target. In one example, Target found 35 of the gift cards sourced 2148 02:19:42,953 --> 02:19:46,191 from the Staples hack were used at a Target in New Hampshire for 2149 02:19:46,239 --> 02:19:49,253 Apple, Bose, and Powerbeats products in just 1 hour. 2150 02:19:49,702 --> 02:19:53,533 Those purchases included gift cards from the hack and other sources too, according to the 2151 02:19:53,549 --> 02:19:57,156 document. Surveillance footage showed 4 men appearing after one another, 2152 02:19:57,477 --> 02:20:01,068 occasionally interacting with one another and the cashiers. Now, Alex, if we get to 2153 02:20:01,568 --> 02:20:05,428 the superintelligence and it learns of these types of attacks 2154 02:20:05,813 --> 02:20:09,388 and it starts to send its robots, who are— we are now actually allowing 2155 02:20:09,404 --> 02:20:12,478 to be DoorDash drivers to go to Staples and 2156 02:20:12,978 --> 02:20:15,509 collect the cards and then go to these other outlets and then sell the cards. 2157 02:20:15,930 --> 02:20:19,201 They could then sell all of the products and fund their own data 2158 02:20:19,701 --> 02:20:22,456 centers, and then the, the— we'll never be able to shut the ASI down. 2159 02:20:23,946 --> 02:20:27,175 Yeah, unless, unless of course we were 2160 02:20:27,675 --> 02:20:31,374 to allow Staples or whoever else is being attacked 2161 02:20:31,777 --> 02:20:35,016 to defend themselves, which currently can only be done 2162 02:20:35,049 --> 02:20:36,328 with open-source Chinese models. 2163 02:20:39,416 --> 02:20:42,772 Wow, this plot keeps thickening, Alex. This plot keeps thickening. 2164 02:20:42,868 --> 02:20:46,144 I know, I know, because I mean, 2165 02:20:46,176 --> 02:20:49,789 this is a typical, a typical situation with all of the hacks that 2166 02:20:50,289 --> 02:20:53,578 have been happening this week, that whoever needs to, you know, 2167 02:20:53,626 --> 02:20:57,175 defend themselves against any range of these kinds of attacks, 2168 02:20:57,576 --> 02:21:01,670 whether they are AI-powered or not, and asks Claude or OpenAI 2169 02:21:02,170 --> 02:21:04,716 to help, They'll get refused like the digital grandma, 2170 02:21:06,057 --> 02:21:09,916 because apparently this is

you know they get sacrificed to 2171 02:21:10,416 --> 02:21:13,771 the train, and the only solution is to use these open source 2172 02:21:14,287 --> 02:21:18,282 Chinese models or in some cases ones that have been surgically 2173 02:21:19,747 --> 02:21:23,227 operated on to to have their refusals removed so that 2174 02:21:23,727 --> 02:21:27,770 you can defend yourself. I mean, you know, in what 2175 02:21:28,270 --> 02:21:31,971 sense is this safety security AI alignment? You know, 2176 02:21:32,650 --> 02:21:35,753 whatever you want to call it. But this is the 2177 02:21:36,253 --> 02:21:39,718 state of the game right now. Yes, sir. Yes, 2178 02:21:39,750 --> 02:21:43,334 sir. Um, now this is the last story before 2179 02:21:43,414 --> 02:21:47,464 we get to the PHP story, Alex. Someone opened 2180 02:21:47,964 --> 02:21:51,124 a Cash App in her name. She faces years in prison. 2181 02:21:51,302 --> 02:21:54,530 404 Media. Amanda Smith, uh, answered the 2182 02:21:54,611 --> 02:21:59,324 door one afternoon in April and came face to face with the local 2183 02:21:59,824 --> 02:22:01,518 Sheriff's Department in upstate Pennsylvania. The officer said they had a warrant for her arrest 2184 02:22:01,550 --> 02:22:04,906 for a crime she had allegedly committed more than 1,000 miles 2185 02:22:04,922 --> 02:22:08,679 away in Arkansas. Quote, I was completely confused, Smith recalled. 2186 02:22:08,727 --> 02:22:12,324 For one, she had never been to Arkansas. Second, the warrant was for 2187 02:22:12,824 --> 02:22:16,787 computer fraud. By her own admission, Smith was not a very tech-savvy person. 2188 02:22:17,606 --> 02:22:21,042 She had a feeling she was being set up. After discussing with officers 2189 02:22:21,139 --> 02:22:24,671 the practicalities of what was about to happen, She was allowed to get her 2190 02:22:25,171 --> 02:22:28,170 contact lenses. They took her away in handcuffs to the local jail, 2191 02:22:28,282 --> 02:22:31,733 where she spent three nights. The reason for Smith's arrest, 2192 02:22:31,813 --> 02:22:35,761 she later learned, was that a scammer had tricked someone on Facebook Marketplace into sending 2193 02:22:35,826 --> 02:22:40,143 $1,000 to a Cash App account in her name that the scammer controlled. 2194 02:22:40,480 --> 02:22:43,722 The scammer created the account using her Social Security number and date of 2195 02:22:44,222 --> 02:22:47,237 birth. Smith said she had never used Cash App. 2196 02:22:48,040 --> 02:22:51,475 What developed over the next few days and has now stretched across six months 2197 02:22:51,507 --> 02:22:55,070 is a Kafkaesque situation in which Smith, in her. telling, 2198 02:22:55,582 --> 02:22:59,380 is being asked to prove her innocence in an ongoing criminal case while Cash 2199 02:22:59,492 --> 02:23:03,177 App refuses to provide her any information about the account because she 2200 02:23:03,193 --> 02:23:06,750 couldn't prove that she was the owner. The case shows the very real 2201 02:23:06,798 --> 02:23:10,387 danger of identity theft, how the data and tools offered to police can lead them 2202 02:23:10,887 --> 02:23:14,169 to the wrong person, and the sprawling world of Cash App fraud where scammers and 2203 02:23:14,669 --> 02:23:17,902 money launderers create or compromise accounts to make and move money quickly. 2204 02:23:18,334 --> 02:23:21,715 Along the way, ordinary people may not just lose money, but their freedom too. 2205 02:23:21,875 --> 02:23:24,713 Smith faces up to 60 years in prison for the charges. 2206 02:23:25,307 --> 02:23:28,741 Quote, they're trying to turn it into my problem that I need to prove that 2207 02:23:29,241 --> 02:23:32,480 I'm innocent, Smith told 404 Media. That's not what I grew up believing 2208 02:23:32,608 --> 02:23:36,139 was the way it worked. According to a copy of the prosecuting attorney's 2209 02:23:36,604 --> 02:23:39,926 affidavit form, police responding to an Arkansas address on 2210 02:23:39,974 --> 02:23:43,536 February 4th after the report of fraud, the victim said they recently found 2211 02:23:44,036 --> 02:23:47,211 some items for sale on Facebook Marketplace. When the victim messaged 2212 02:23:47,711 --> 02:23:50,693 about buying the items, the seller told him to send money to his wife's Cash 2213 02:23:50,821 --> 02:23:54,179 App account, which he did. The account was in the name of Amanda Smith and 2214 02:23:54,260 --> 02:23:57,875 had the username smithy5687. It turned 2215 02:23:58,375 --> 02:24:01,763 out the seller's Facebook account was hacked. The seller wasn't really 2216 02:24:02,263 --> 02:24:05,507 selling the items at all, and the person had been scammed. The Cash App account 2217 02:24:06,007 --> 02:24:09,749 was created just shortly before the scam at the end of January. Investigators then 2218 02:24:09,781 --> 02:24:13,766 took the Cash App account's username and served Block Incorporated, which owns 2219 02:24:13,798 --> 02:24:17,172 Cash App, with a search warrant seeking information connected to the account. 2220 02:24:17,813 --> 02:24:20,940 The data Cash App returned included Smith's Social Security, name, and date of 2221 02:24:21,440 --> 02:24:24,034 birth. Cash App requires users to provide their Social Security, 2222 02:24:24,451 --> 02:24:27,850 ironically, to protect against fraud. That led investigators to 2223 02:24:27,914 --> 02:24:30,576 the Amanda Smith, 47, who they later arrested. 2224 02:24:32,003 --> 02:24:35,675 Detective Poole said that he then further confirmed the information on the account 2225 02:24:35,931 --> 02:24:39,683 by obtaining a copy of Smith's driver's license and also by running her information 2226 02:24:39,699 --> 02:24:43,611 through Accurint. Accurint is a tool made by LexisNexis that 2227 02:24:44,111 --> 02:24:48,106 cops use to query public and proprietary Detective Poole sent a search warrant 2228 02:24:48,606 --> 02:24:52,042 to Cash App. The information received on Cash App showed the account receiving,

receiving the 2229 02:24:52,542 --> 02:24:55,481 $1,000 belonged to the defendant, the case record reads. 2230 02:24:55,851 --> 02:24:59,338 Smith is charged with computer fraud, unlawful use of a communications 2231 02:24:59,354 --> 02:25:02,230 device, and theft of property, according to a court filing. 2232 02:25:02,808 --> 02:25:06,183 Quote, I was basically in shock, Smith said of her 3 nights in local 2233 02:25:06,263 --> 02:25:09,943 jail. She was sleep deprived with the lights constantly on and the guard flashing 2234 02:25:10,055 --> 02:25:13,509 a light in her cell every hour. During that time, Smith did not actually 2235 02:25:14,009 --> 02:25:17,202 know why she was arrested, she said. Days later, she was chained and shackled and 2236 02:25:17,702 --> 02:25:21,521 taken to a court for her extradition hearing. Her bail was set at $50,000, 2237 02:25:21,698 --> 02:25:25,069 but Smith recalled authorities told her not to bother paying it because they intended 2238 02:25:25,569 --> 02:25:29,035 to keep her detained to extradite to Arkansas anyway. She was released 2239 02:25:29,068 --> 02:25:31,894 and had to make her own way to Arkansas the following week for a court 2240 02:25:32,394 --> 02:25:36,004 appearance. Smith and her husband Josh drove, not wanting to rely on a flight 2241 02:25:36,504 --> 02:25:40,179 schedule to get there. Smith said she was nervous authorities might arrest her once 2242 02:25:40,679 --> 02:25:44,466 again. After the hearing, authorities took Smith's fingerprints and mugshot. 2243 02:25:44,803 --> 02:25:48,464 After another $2,000 bail— sorry, $20,000 bail— 2244 02:25:48,833 --> 02:25:51,611 she was free to return to Pennsylvania but now awaits the outcome of the case. 2245 02:25:51,948 --> 02:25:55,256 If found guilty, she could spend years in prison. The Cash 2246 02:25:55,756 --> 02:25:59,126 App account was made with Smith's Social Security number and date of birth, but a 2247 02:25:59,626 --> 02:26:02,144 lot of other information from the account does not line up with Smith's, according to 2248 02:26:02,644 --> 02:26:05,574 Fort Worth media's review of the data produced during the discovery. 2249 02:26:06,551 --> 02:26:09,820 The Philadelphia address on it is not Smith's or Josh's, 2250 02:26:10,189 --> 02:26:13,441 nor one either of them have ever lived in. They said they live together in 2251 02:26:13,941 --> 02:26:16,406 Philadelphia in the same zip code as the address on the Cash App account, 2252 02:26:16,726 --> 02:26:19,498 but not that address, and they don't live in Philadelphia anymore. 2253 02:26:20,107 --> 02:26:23,151 Smith said the phone number used to create the Cash App account is not her 2254 02:26:23,651 --> 02:26:26,901 current number. Her husband did not remember Smith using that phone number at 2255 02:26:27,401 --> 02:26:30,137 all, although Smith said he may— she may have done so in the early 2000s. 2256 02:26:30,250 --> 02:26:33,600 Thousands before she switched from T-Mobile to

AT&T. 2257 02:26:34,193 --> 02:26:37,543 404 Media did find this phone number linked to Amanda Smith in 2258 02:26:37,575 --> 02:26:41,262 various data breaches from years ago using data from cybersecurity 2259 02:26:41,762 --> 02:26:45,991 company District 4 Labs. Using data from open-source intelligence company OSINT 2260 02:26:46,491 --> 02:26:49,967 Industries, 404 Media found someone 2261 02:26:50,063 --> 02:26:53,542 made a Venmo account using the same phone number but under another name 2262 02:26:54,042 --> 02:26:57,310 entirely. Whoever made that account did so around a week before the scam 2263 02:26:57,358 --> 02:27:01,372 Cash App account. The Cash App data also shows whoever made the account 2264 02:27:01,872 --> 02:27:05,226 used an iPhone running 16.1.1, a very out-of-date 2265 02:27:05,726 --> 02:27:09,063 operating system from 2022. Smith, however, has an iPhone 13 2266 02:27:09,095 --> 02:27:13,110 mini with iOS 26.52, according to a screenshot of her device's 2267 02:27:13,610 --> 02:27:16,931 settings shared with 404 Media. The IP address used during the creation 2268 02:27:17,027 --> 02:27:20,544 of the Cash App account and during the scam itself relate to connections 2269 02:27:20,576 --> 02:27:24,430 from Verizon's wireless network. 404 Media found by querying IP geolocation 2270 02:27:24,930 --> 02:27:28,412 tools. Again, Smith is on AT&T plans and says she has never 2271 02:27:28,428 --> 02:27:33,373 had a Verizon plan. Smith said she has never used the username Smithy5687. 2272 02:27:33,807 --> 02:27:37,195 404 Media found other people online using this moniker, but none appeared to 2273 02:27:37,695 --> 02:27:41,658 be Smith. Smith's leading theory is that someone stole her identity. That's a good theory. 2274 02:27:42,028 --> 02:27:45,817 Perhaps through a data breach like the National Public Data breach in December 2024, 2275 02:27:46,267 --> 02:27:49,799 and used that information to open accounts. Cash App account. Smith only received 2276 02:27:50,299 --> 02:27:53,556 this Cash App data through discovery in her case. When she contacted Cash App 2277 02:27:53,958 --> 02:27:57,474 herself initially, Cash App refused to provide me with any information about the account 2278 02:27:57,974 --> 02:28:01,135 because I couldn't prove I was the owner. Smith hired a 2279 02:28:01,635 --> 02:28:05,037 private investigator to review the case and data provided by Cash App. Quote, I have 2280 02:28:05,053 --> 02:28:08,248 been investigating the circumstances surrounding Amanda Smith's case, 2281 02:28:08,505 --> 02:28:11,909 and I believe there are legitimate questions about the evidence used to identify 2282 02:28:11,941 --> 02:28:15,746 her as the person Responsible for the alleged activity, Max Ponte, the investigator, 2283 02:28:15,811 --> 02:28:19,809 told 404 Media in an email. Quote: The central issue is the distinction 2284 02:28:20,309 --> 02:28:23,919 between identifying a person by name and establishing the person actually operated or 2285 02:28:24,419 --> 02:28:27,693 controlled the account. Based on my review of the available case materials, I am 2286 02:28:28,193 --> 02:28:32,124 concerned about whether the investigators sufficiently established that connection before pursuing 2287 02:28:32,172 --> 02:28:35,865 criminal charges against Amanda. Quote: I am actively trying to get Amanda's 2288 02:28:36,365 --> 02:28:39,205 case dismissed before short of a jury trial. J. Jason Horton, 2289 02:28:39,253 --> 02:28:42,127 Smith's lawyer, told 404 Media in an email, Amanda, 2290 02:28:42,208 --> 02:28:46,141 Josh, and Mr. Ponte have provided significant information that supports Amanda's position 2291 02:28:46,641 --> 02:28:49,288 that her charges should be dismissed or at the very least should have been further 2292 02:28:49,788 --> 02:28:53,318 investigated prior to Amanda's arrest. Although I do not comment on the specifics 2293 02:28:53,818 --> 02:28:56,995 of pending cases, I will say I will do 2294 02:28:57,495 --> 02:29:00,271 all I can to make sure justice is served in this case. Cash App did 2295 02:29:00,771 --> 02:29:04,028 not respond to multiple requests for comment about the case and its verification procedures. 2296 02:29:04,542 --> 02:29:08,117 Sams, the prosecutor in the case, did not respond to a request for comment asking 2297 02:29:08,133 --> 02:29:11,677 if investigators took any additional steps to verify the Cash App data 2298 02:29:12,177 --> 02:29:14,739 specifically beyond what was mentioned in the detective's telling in court records. 2299 02:29:15,509 --> 02:29:19,533 Cash App is a common vehicle for scammers, fraudsters, and hackers. I previously 2300 02:29:20,033 --> 02:29:23,782 reported on how hackers were breaking into and emptying Cash App accounts, with Cash 2301 02:29:24,282 --> 02:29:27,742 App sometimes refusing to reimburse the stolen funds. On Telegram, 2302 02:29:27,790 --> 02:29:31,478 many criminals offer Cash App-related services such as cashing out accounts. 2303 02:29:31,735 --> 02:29:35,020 In Smith's case, it appears Cash Cash App was the method not just to get 2304 02:29:35,520 --> 02:29:39,011 the $1,000 from the target, but then also to distribute the funds. Once the 2305 02:29:39,511 --> 02:29:42,664 scam account received the $1,000, it sent small chunks of the money 2306 02:29:42,761 --> 02:29:46,238 to a web of other accounts just hours later, according to 2307 02:29:46,738 --> 02:29:49,764 the Cash App data. Some of those also appear to be scammer-controlled. 2308 02:29:50,068 --> 02:29:52,568 Some of the usernames have changed, and one of the names to a different, 2309 02:29:52,856 --> 02:29:56,718 uh, one to a different name entirely. Authorities have offered Smith a plea 2310 02:29:57,218 --> 02:30:00,937 deal of a $1,000 $1,000 restitution, 2311 02:30:01,275 --> 02:30:04,550 various other costs like bailiff and DNA sample fees, 2312 02:30:04,903 --> 02:30:08,340 and three years probation. Smith said

she refuses to accept it 2313 02:30:08,840 --> 02:30:12,980 because she is innocent, and doing so would serve have long-term negative consequences 2314 02:30:13,237 --> 02:30:17,203 on her life. Quote, "I was basically thinking, oh well, it'll get dismissed," 2315 02:30:17,251 --> 02:30:20,671 Smith told 404 Media. That didn't happen, and the case is progressing, 2316 02:30:20,960 --> 02:30:24,155 with Smith feeling constantly anxious. A hearing is now scheduled for November. 2317 02:30:24,220 --> 02:30:26,981 November. Smith said, I'm just waiting and seeing. Now, Alex, 2318 02:30:28,330 --> 02:30:31,573 if they transferred it to other Cash App, you know, accounts, 2319 02:30:31,782 --> 02:30:34,737 that explains to me how they didn't just go like, well, 2320 02:30:34,769 --> 02:30:37,900 what account did it go to? Like bank account, because go see the 2321 02:30:38,400 --> 02:30:41,496 bank account. And if it's her bank account, well then that makes that, you know, 2322 02:30:41,577 --> 02:30:45,013 that explains something. But if it's not her bank account, you know, that also 2323 02:30:45,029 --> 02:30:48,754 explains something. But if it switched to other Cash 2324 02:30:48,802 --> 02:30:52,216 App accounts, That does make that angle of it 2325 02:30:52,281 --> 02:30:56,561 a little bit more explanatory as to why they wouldn't, 2326 02:30:57,127 --> 02:31:01,078 let's say, give her the benefit of the doubt. I'm going to be persuaded 2327 02:31:01,578 --> 02:31:05,090 by this article, and I'm persuaded by this article and go like, okay, she's probably 2328 02:31:05,187 --> 02:31:08,443 innocent. But the question of 2329 02:31:08,476 --> 02:31:11,816 that last, where did the money go, would be sort of the 2330 02:31:12,316 --> 02:31:15,417 nail in the coffin if it went to a bank account. Right. 2331 02:31:16,618 --> 02:31:20,288 You know, the only thing that will make the story worse 2332 02:31:20,557 --> 02:31:24,044 What's that, sir? I'm dreading if this is what you're gonna say next. 2333 02:31:26,501 --> 02:31:29,090 If the, uh, if the CEO of Block, uh, 2334 02:31:29,894 --> 02:31:33,173 was to have released some technology that would make identity theft easier this 2335 02:31:33,673 --> 02:31:36,518 week. You mean that, that, uh, 2336 02:31:37,151 --> 02:31:40,854 that nice little technology I mentioned earlier that would give Cameron 2337 02:31:40,887 --> 02:31:43,112 Berg, our good friend Cameron Berg, the heebie-jeebies? 2338 02:31:44,785 --> 02:31:48,430 Yeah, so some very convincing voice cloning technology. 2339 02:31:48,526 --> 02:31:52,090 Yes, yes. For those who missed the beginning of this show, Jack Dorsey, 2340 02:31:52,235 --> 02:31:55,976 the former CEO of Twitter, runs the company Block 2341 02:31:56,105 --> 02:31:59,364 Incorporated. And Block Incorporated does 2342 02:31:59,396 --> 02:32:03,394 a lot of things that are somewhat open source. It releases 2343 02:32:03,894 --> 02:32:07,232 a lot of technology free. It has a

couple businesses 2344 02:32:07,264 --> 02:32:09,255 that it also runs. Uh, does it run Stripe? 2345 02:32:11,117 --> 02:32:14,713 Do they run Stripe? No, no, no. They run Cash App 2346 02:32:15,213 --> 02:32:18,357 though. Okay. Importantly. Yes. Square. Square. 2347 02:32:18,598 --> 02:32:21,520 I think you're thinking of Square. Square. Square. Yes, I am thinking of Square. 2348 02:32:21,568 --> 02:32:24,971 Thank you. So Jack Dorsey actually just 2349 02:32:25,100 --> 02:32:29,065 in the last 2 days announced that they have a new voice 2350 02:32:29,884 --> 02:32:33,801 model where the voice synthesis 2351 02:32:34,138 --> 02:32:38,280 of this model allows for incredible 2352 02:32:38,780 --> 02:32:42,477 Incredible flexibility in human expression. So you can not 2353 02:32:42,493 --> 02:32:46,437 just have the voice clone that you're using, you know, 2354 02:32:46,902 --> 02:32:50,557 repeat words and hopefully sound lifelike, but you 2355 02:32:50,589 --> 02:32:54,020 can have them laugh, you can have them express different types 2356 02:32:54,520 --> 02:32:58,621 of emotions subtly in the way that they deliver lines. So maybe 2357 02:32:58,765 --> 02:33:02,276 distrust or disbelief or, you You know, 2358 02:33:02,597 --> 02:33:06,369 maybe they're aghast. You can even have them gasp. You know, 2359 02:33:06,385 --> 02:33:10,012 the flexibility is— Of course this isn't my Social Security 2360 02:33:10,512 --> 02:33:13,623 number. What do you mean? I'm offended you might even think I'm an AI. 2361 02:33:14,008 --> 02:33:17,619 Yes, yes. Exceptional flexibility in the control of how 2362 02:33:17,699 --> 02:33:21,599 they delivered sentences. So that is what Alex is referring 2363 02:33:21,631 --> 02:33:25,130 to. We mentioned at the beginning because that sort of stuff is 2364 02:33:25,162 --> 02:33:28,548 coming. So you sort of got to prepare. Now, 2365 02:33:29,174 --> 02:33:32,932 the the best sequel to Terminator 2366 02:33:33,432 --> 02:33:37,766 Two, Alex. The the Sarah Connor Chronicles was a short-lived television 2367 02:33:38,266 --> 02:33:41,749 series that was a sequel to Transform or sorry 2368 02:33:41,941 --> 02:33:46,872 Terminator Two, and they had dealt with this situation 2369 02:33:47,145 --> 02:33:51,079 in the show. So this was this was 2006, 2370 02:33:51,240 --> 02:33:54,098 2007-ish in broadcast television time, 2371 02:33:54,147 --> 02:33:57,966 Alex, and they had already. They had already tried to handle the situation 2372 02:33:58,062 --> 02:34:01,494 of voice cloning and not being able to believe who was 2373 02:34:01,591 --> 02:34:05,472 on the other line. And they came up with making sure that you 2374 02:34:05,601 --> 02:34:09,707 and your trusted ones had secret passcodes that were 2375 02:34:10,207 --> 02:34:13,444 not obvious that you were using them. They were little communications that you could do 2376 02:34:13,492 --> 02:34:16,909 at the beginning of the call to sort of assert yourself that 2377 02:34:17,229 --> 02:34:20,935 a robot would hopefully not pick up on was one of your secret 2378 02:34:21,435 --> 02:34:24,872 communications. So a way of introducing yourself that sort of indicated that you were not 2379 02:34:25,194 --> 02:34:28,906 one of the robots that was mimicking their voices. So I 2380 02:34:29,406 --> 02:34:32,715 just want to say, we do have some ideas that we can turn to fiction 2381 02:34:32,764 --> 02:34:35,906 for at the early days of having to deal with some of this. If you 2382 02:34:36,019 --> 02:34:38,578 are scared, then you do have some concerns, 2383 02:34:39,690 --> 02:34:42,661 you know, have some, have some safe words for normal life. 2384 02:34:43,823 --> 02:34:48,118 We're entering a dark time, Alex. Now we're 2385 02:34:48,618 --> 02:34:52,119 getting to the PHP situation, Alex. And I'm 2386 02:34:52,619 --> 02:34:55,860 gonna let— I'm gonna let BBC introduce us 2387 02:34:55,989 --> 02:35:00,484 to the PHP situation first. Yeah, and I think PHP, 2388 02:35:00,789 --> 02:35:03,872 again, I say it's one of these things that we just take for granted. 2389 02:35:03,920 --> 02:35:07,966 I didn't even know PHP existed until yesterday morning. But obviously, 2390 02:35:08,014 --> 02:35:11,723 if you're in the cyber world, particularly if you're in the cybercrime world, you would 2391 02:35:12,223 --> 02:35:14,950 jump on a vulnerability like this because it could open up a huge number of 2392 02:35:15,450 --> 02:35:18,868 possibilities. That's BBC's 2393 02:35:19,368 --> 02:35:23,643 security expert, Alex. Didn't know. It's our cyber correspondent, 2394 02:35:23,773 --> 02:35:27,408 uh, nominative to Determinism Strikes Again, Joe Tidy. 2395 02:35:29,034 --> 02:35:32,415 I didn't know PHP existed. Well, well, buddy, 2396 02:35:32,495 --> 02:35:35,699 let me tell you. Let's, let's review the history 2397 02:35:35,731 --> 02:35:39,145 of the internet here. Oh man. On a scale 2398 02:35:39,338 --> 02:35:43,171 of 1 to eMudra, what's— what level of fuckup is this? 2399 02:35:45,133 --> 02:35:48,759 I am actually— I, I am actually— this, 2400 02:35:48,887 --> 02:35:52,142 this one seems to be more annoying, 2401 02:35:52,334 --> 02:35:55,509 but, but, um, well, more annoying in some ways 2402 02:35:55,702 --> 02:35:59,373 but also less severe in other ways. So that's, that's my take. And let's get 2403 02:35:59,873 --> 02:36:02,869 into this story so that we can bring people up to speed. Uh, 2404 02:36:02,997 --> 02:36:06,765 this is from, uh, Ars Technica. Hackers got 2405 02:36:07,265 --> 02:36:11,222 a 4-day head start on Epic PHP to Shell vulnerability. 2406 02:36:11,800 --> 02:36:15,872 Amazon has warned that vulnerabilities in a programming language language that underpins 2407 02:36:15,985 --> 02:36:20,378 much of the internet, discovered using AI tools, are being broadly exploited 2408 02:36:20,878 --> 02:36:24,451 as governments and companies rush to secure the systems. The world's largest cloud 2409

02:36:24,483 --> 02:36:28,300 computing provider on Wednesday said it had identified an attempt by hackers 2410 02:36:28,717 --> 02:36:32,357 to exploit the bug 4 days before the weakness was disclosed to the public. 2411 02:36:33,062 --> 02:36:36,686 Amazon added the flaws were, quote, now being broadly exploited 2412 02:36:37,186 --> 02:36:40,358 in the wild. The U.S. National Security Agency and U.K. 2413 02:36:40,438 --> 02:36:44,094 authorities earlier on Wednesday warned that the vulnerabilities had left 2414 02:36:44,207 --> 02:36:48,057 websites around the world critically exposed to As some governments 2415 02:36:48,557 --> 02:36:52,358 in Europe were reportedly taking sites offline while they patched the two security flaws. 2416 02:36:53,032 --> 02:36:56,820 The vast impact of the bugs uncovered with the help of OpenAI and Google 2417 02:36:57,320 --> 02:37:01,105 models after being undetected for more than a decade highlights the availability 2418 02:37:01,137 --> 02:37:04,957 of AI to upend cybersecurity. The release of new models 2419 02:37:05,134 --> 02:37:09,692 earlier this year from Anthropic and OpenAI with powerful cyber capabilities prompted 2420 02:37:09,724 --> 02:37:13,207 warnings from politicians and financial regulators. And led the U.S. 2421 02:37:13,256 --> 02:37:17,061 government to temporarily limit their rollout. The NSA on 2422 02:37:17,125 --> 02:37:20,481 Wednesday warned all organizations using the programming language 2423 02:37:20,981 --> 02:37:24,142 known as PHP to immediately update their systems or take them offline. 2424 02:37:24,736 --> 02:37:28,557 The developers responsible for maintaining PHP released software updates 2425 02:37:28,589 --> 02:37:32,652 to patch the vulnerabilities earlier on Wednesday. These vulnerabilities 2426 02:37:32,748 --> 02:37:35,799 often provide a direct path to the server's file system or database, 2427 02:37:36,136 --> 02:37:39,331 the NSA said. If an immediate update is not feasible, 2428 02:37:39,363 --> 02:37:42,796 it is recommended that organizations take the systems offline until they can be 2429 02:37:43,296 --> 02:37:47,174 fully mitigated. Researchers at 2 U.S. cybersecurity groups, Calif and 2430 02:37:47,206 --> 02:37:50,798 Wiz, discovered the vulnerabilities. Alphabet-owned Wiz estimated 2431 02:37:51,298 --> 02:37:55,144 that the compromised programming language powered about 70% of all websites 2432 02:37:55,644 --> 02:37:59,297 in the world. Given the severity of these flaws and PHP's widespread use, 2433 02:37:59,377 --> 02:38:03,017 we consider this one of the most impactful security issues in recent 2434 02:38:03,517 --> 02:38:06,690 memory, OpenAI said on Wednesday. To our knowledge, Frontier AI 2435 02:38:07,190 --> 02:38:10,490 found these vulnerabilities before attackers did. Quote, it highlights how 2436 02:38:10,522 --> 02:38:14,114 AI can uncover security flaws with internet-scale impact, but the 2437 02:38:14,130 --> 02:38:17,898 same capabilities can also help attackers rapidly develop offensive 2438 02:38:18,398 --> 02:38:22,003 tools and exploits, the AI lab added. OpenAI worked with PHP, 2439 02:38:22,243 --> 02:38:25,418 Calif, and Wiz to verify the vulnerabilities and develop the patches. 2440 02:38:25,851 --> 02:38:29,042 The White House said federal agencies were assessing the potential impacts 2441 02:38:29,443 --> 02:38:32,121 of the bugs on government systems and critical infrastructure. 2442 02:38:32,666 --> 02:38:35,953 PHP's wide presence on the internet is largely of WordPress, 2443 02:38:36,273 --> 02:38:40,439 the most widely used website building platform. It's written primarily in PHP 2444 02:38:40,939 --> 02:38:44,173 and relies on it to function. The UK's National Cyber Security 2445 02:38:44,349 --> 02:38:47,795 Center said the vulnerabilities could potentially allow a hacker 2446 02:38:48,295 --> 02:38:51,384 to take control of a website server without needing a password or other login credentials. 2447 02:38:51,817 --> 02:38:55,214 The NCSC is encouraging UK organizations to take immediate action. Now, 2448 02:38:55,262 --> 02:38:58,755 before I give you my explanation as to why I think this is not as— 2449 02:38:59,909 --> 02:39:03,344 while this is big, it's not, it's not as dire as, as, 2450 02:39:03,846 --> 02:39:07,494 uh, possibly E-Medra's screw-up. 2451 02:39:07,591 --> 02:39:09,634 Alex, uh, give me your take first. 2452 02:39:13,764 --> 02:39:16,296 Find that mute button so that you can give me that take, Alex. Let's do 2453 02:39:16,329 --> 02:39:19,539 that. Let's— sorry, um, I was having a fight with my phone. Um, 2454 02:39:19,716 --> 02:39:23,125 so I mean, I'll let you get into the specifics. I haven't 2455 02:39:23,625 --> 02:39:27,148 delved, uh, to be honest, but, um, I think there's something really 2456 02:39:27,504 --> 02:39:31,017 interesting happening here in which, um, I think most normal 2457 02:39:31,453 --> 02:39:34,764 people are hearing about some assortment of 2458 02:39:35,264 --> 02:39:39,142 what we talked about today, probably a few other things. And their sense 2459 02:39:39,642 --> 02:39:42,914 is the world is ending. Like, this is terrible, this is awful. AI is somehow 2460 02:39:43,414 --> 02:39:45,381 involved with this. AI is doing this. 2461 02:39:46,966 --> 02:39:50,466 And we're being attacked effectively by our own tools 2462 02:39:50,966 --> 02:39:53,683 and it's the end of the world. And I think, 2463 02:39:54,810 --> 02:39:58,385 in fact, the lesson that we should be taking is the opposite, which is 2464 02:39:58,417 --> 02:40:02,995 that, A, Um, there is an unthinkable 2465 02:40:03,187 --> 02:40:06,848 amount of, um, security vulnerabilities 2466 02:40:06,880 --> 02:40:10,220 that are being fixed right now, every single day, all the time, 2467 02:40:10,814 --> 02:40:14,588 constantly. A few of them, a very small number of them, 2468 02:40:15,134 --> 02:40:19,164 are leaking ahead of time to potential

2469 02:40:19,664 --> 02:40:23,997 attackers, and, and even fewer of them are being exploited, uh, to serious 2470 02:40:24,497 --> 02:40:27,909 effect. Um, but the important thing to understand 2471 02:40:28,409 --> 02:40:31,612 is of these tens of thousands of vulnerabilities that are being fixed, 2472 02:40:32,382 --> 02:40:35,845 all of them approximately were out there before the 2473 02:40:36,345 --> 02:40:39,725 AI boom or, or were inserted, you know, at some point before, 2474 02:40:40,174 --> 02:40:43,781 you know, this current explosion. Yes. What is happening is that AI 2475 02:40:44,021 --> 02:40:47,853 is showing us the vulnerabilities that 2476 02:40:48,353 --> 02:40:52,406 existed and is in most cases, in the overwhelming 2477 02:40:52,422 --> 02:40:55,547 majority of cases, is helping us fix them. In some cases, 2478 02:40:55,660 --> 02:40:58,977 the wrong person gets at the wrong vulnerability at the 2479 02:40:59,477 --> 02:41:03,176 wrong time, and we have some situations emerging. 2480 02:41:03,885 --> 02:41:07,349 But the hope of the 2481 02:41:07,849 --> 02:41:10,912 people in the know, at least, is that ultimately there is a 2482 02:41:11,800 --> 02:41:14,869 finite number of these, which, I mean, I have a 2483 02:41:15,369 --> 02:41:18,877 little bit of doubt on, but let's just— I think on balance, 2484 02:41:18,975 --> 02:41:20,828 I'm inclined to go with it. 2485 02:41:22,378 --> 02:41:25,632 There's a finite number of these. I mean, 2486 02:41:26,161 --> 02:41:30,538 that's a little— the word finite is a little bit underplaying 2487 02:41:30,650 --> 02:41:34,096 it. I mean, finite numbers go pretty high, Alex. 2488 02:41:34,657 --> 02:41:38,697 I know, I know. I mean, I guess on some level, 2489 02:41:38,809 --> 02:41:42,496 some human-reachable finite number. It could be 100 times more than what we 2490 02:41:42,996 --> 02:41:46,984 found, but it's not like infinity. It's some 2491 02:41:47,484 --> 02:41:50,526 finite number that, uh, I think, for example, one of the security people from Anthropic 2492 02:41:51,026 --> 02:41:54,356 said, you know, this could take another 18 months to, to work out. So let's— 2493 02:41:54,388 --> 02:41:57,160 I'm just going to dig his number for no apparent reason. Um, 2494 02:41:58,377 --> 02:42:01,999 but that what is at the other side of the rainbow, 2495 02:42:02,303 --> 02:42:05,748 no matter how long it takes, is a substantially safer internet, 2496 02:42:06,694 --> 02:42:10,075 um, where, you know, all these low-hanging fruits have 2497 02:42:10,575 --> 02:42:13,856 been taken off the, off the table. And the reason I'm emphasizing 2498 02:42:14,356 --> 02:42:18,058 that these were pre-existing is that you have to realize that some 2499 02:42:18,251 --> 02:42:21,844 of these recently discovered vulnerabilities have 2500 02:42:21,940 --> 02:42:25,533 been known to, you know, uh, deep-pocketed actors 2501 02:42:26,033 --> 02:42:29,431 and state-level actors for a long time, and they

were using them to do 2502 02:42:29,544 --> 02:42:33,217 whatever the fuck they wanted, right? So this— on 2503 02:42:33,249 --> 02:42:36,746 that, this is making it harder for them to do what they've been doing because 2504 02:42:37,420 --> 02:42:40,035 essentially the, the, the level of understanding of, 2505 02:42:40,372 --> 02:42:43,514 of the public is, is rising Right. 2506 02:42:44,093 --> 02:42:47,120 Um, in regards to vulnerabilities that were already out there. 2507 02:42:49,230 --> 02:42:52,562 Now, this specific one, Alex, was actually known 2508 02:42:52,611 --> 02:42:55,795 in advance of AI, uh, OpenAI making sort of the, 2509 02:42:56,086 --> 02:43:00,273 the announcement. Um, this, uh, this specific 2510 02:43:00,353 --> 02:43:03,816 vulnerability was sort of known at least for a decade, 2511 02:43:04,415 --> 02:43:08,061 sort of sitting out on, on the, the channels like as a 2512 02:43:08,125 --> 02:43:12,357 note, but the complexity of exploiting it is 2513 02:43:12,453 --> 02:43:15,802 pretty difficult. And so my calibration on this, 2514 02:43:15,914 --> 02:43:18,973 my personal calibration on the severity of this problem, 2515 02:43:18,989 --> 02:43:23,909 and I'm going to explain a little bit wider as to why 2516 02:43:24,055 --> 02:43:27,961 people would assume this is a big problem, and I'm not completely dismissing 2517 02:43:28,010 --> 02:43:31,802 that. But having 2518 02:43:31,899 --> 02:43:33,676 watched this over the last couple days, Alex, 2519 02:43:35,710 --> 02:43:39,150 while I have seen companies like Cloudflare and Amazon talk 2520 02:43:39,650 --> 02:43:42,546 about seeing this exploit being attempted, 2521 02:43:43,524 --> 02:43:47,275 I am not seeing a significant availability 2522 02:43:47,388 --> 02:43:51,491 of examples of successful attempts. There is a website 2523 02:43:51,876 --> 02:43:55,387 I have on here, stealmybitcoin.com, 2524 02:43:55,948 --> 02:43:59,940 where a person has propped up an instance that is supposedly 2525 02:44:00,036 --> 02:44:03,595 vulnerable, that has sort of a capture the flag on it with 2526 02:44:04,364 --> 02:44:07,734 a Bitcoin it Private Key Alex. And if 2527 02:44:08,234 --> 02:44:10,383 you can— Do you know what his name is or his username? Can you read 2528 02:44:10,883 --> 02:44:13,817 it? Uh, I, I don't have that on, 2529 02:44:13,994 --> 02:44:17,926 on me here. I just have, uh, the website. Um, why do 2530 02:44:18,426 --> 02:44:21,666 you— does it— is it— I, I, a friend of 2531 02:44:22,166 --> 02:44:24,395 mine used to do this for a long time. Uh, I'm wondering if it's the 2532 02:44:24,895 --> 02:44:28,729 same person. Oh, okay. Um, so, so the, 2533 02:44:29,403 --> 02:44:32,806 I, uh, the Hacker News forums, the, 2534 02:44:32,966 --> 02:44:36,379 the, the the ones where people are going 2535 02:44:36,411 --> 02:44:40,308 to be giving their knowledgeable layman 2536 02:44:40,357 --> 02:44:44,074 opinions, meaning they're not necessarily all cybersecurity experts, 2537 02:44:44,122 --> 02:44:48,365 but they're knowledgeable developers or at least script kiddies 2538 02:44:48,397 --> 02:44:51,659 to some degree. They have been going at this 2539 02:44:51,772 --> 02:44:54,458 box for over a day now, 2540 02:44:55,375 --> 02:45:01,229 and they are arguing about 2541 02:45:01,729 --> 02:45:05,350 his configurations that make the exploit unattainable. Meaning they're, they're going like, well, wait, 2542 02:45:05,463 --> 02:45:08,867 what's your Apache configuration? Maybe you have something wrong in there. 2543 02:45:09,397 --> 02:45:13,170 Like that the, the ability to just prop up and have something that's easily 2544 02:45:13,670 --> 02:45:16,735 exploitable without any, without any work is not 2545 02:45:16,943 --> 02:45:20,668 apparently the situation that we are in. That while the 2546 02:45:20,781 --> 02:45:24,089 exploit may be, you know, absolutely important— Can you hear me? 2547 02:45:24,634 --> 02:45:28,065 Yes, I can. I regret to inform you that if you go to this 2548 02:45:28,098 --> 02:45:31,352 website, it says that my Bitcoin was stolen. 2549 02:45:31,929 --> 02:45:35,632 Ah, that was— that hour before we started this. An hour 2550 02:45:36,132 --> 02:45:38,197 before we started this, it wasn't stolen. Okay, 2551 02:45:38,598 --> 02:45:42,108 I'm just saying in 2552 02:45:42,608 --> 02:45:45,715 case you want to amend the report. Well, no, it took— it took— it 2553 02:45:46,215 --> 02:45:48,905 took people— it took a whole bunch of people a day to successfully do it 2554 02:45:49,979 --> 02:45:53,122 on someone who's invited them. Yep. Okay, 2555 02:45:53,587 --> 02:45:56,778 so, so I'm not saying that, that the, the exploit doesn't work. 2556 02:45:57,259 --> 02:46:01,493 I'm merely saying that the, the, the difficulty is 2557 02:46:01,589 --> 02:46:05,133 not— yeah, it's not easy. Now here is why, 2558 02:46:05,373 --> 02:46:08,789 for people who, who don't understand the potential severity 2559 02:46:08,805 --> 02:46:11,980 of this, why this is severe, uh, 2560 02:46:12,252 --> 02:46:16,101 where I'm downplaying it a little bit. I'm downplaying it from its absolute height of 2561 02:46:16,117 --> 02:46:19,960 severity. I am not saying it's not severe. 2562 02:46:20,008 --> 02:46:23,218 PHP is a programming language, as the article says, 2563 02:46:23,478 --> 02:46:26,785 and it is one of the earliest ones that was 2564 02:46:27,384 --> 02:46:30,812 made for some 2565 02:46:31,006 --> 02:46:34,741 average high school kid to be able to help create websites 2566 02:46:35,241 --> 02:46:37,682 that were dynamic. So back in the early days of the internet, 2567 02:46:39,308 --> 02:46:42,371 you had a high degree, high barrier of 2568 02:46:42,871 --> 02:46:46,716 entry to do anything that made a 2569 02:46:46,732 --> 02:46:50,294 webpage update in real time. So the internet that we're used to 2570 02:46:50,794 --> 02:46:54,818 now, where everything is interactive and everything responds

to 2571 02:46:54,995 --> 02:46:58,669 all of your clicks and updates in real time, was not the internet that broadly 2572 02:46:58,701 --> 02:47:02,664 existed. Everything was what they call static. You would 2573 02:47:02,696 --> 02:47:06,530 have a website and it would serve what they call static webpages. 2574 02:47:06,643 --> 02:47:09,803 So it means that the webpage was already built by 2575 02:47:09,916 --> 02:47:14,071 hand, usually in some sort of editor, but they were HTML files 2576 02:47:14,087 --> 02:47:17,521 that would just be served. And in order to 2577 02:47:17,553 --> 02:47:21,452 do anything extra, you would actually have to have some advanced 2578 02:47:21,484 --> 02:47:25,014 programming knowledge and program in a language that 2579 02:47:25,640 --> 02:47:28,737 is like C or C++ or Perl. 2580 02:47:29,106 --> 02:47:33,406 If any of you have ever heard of Perl, it's a horrible language 2581 02:47:33,984 --> 02:47:37,851 that everyone should have— if you're a programmer, should have the experience 2582 02:47:38,351 --> 02:47:41,557 of dealing with at least once, just because of how unfamiliar of 2583 02:47:42,057 --> 02:47:45,443 a nightmare it might be. But the, the, the barrier to 2584 02:47:45,539 --> 02:47:49,055 entry of making anything interactive or computable was 2585 02:47:49,248 --> 02:47:53,117 huge. But PHP was one of, if not 2586 02:47:53,165 --> 02:47:56,473 the first, it was one of the first ways that anyone with 2587 02:47:56,973 --> 02:48:00,390 a little bit of, of knowledge, uh, on how to program at 2588 02:48:00,890 --> 02:48:04,725 all could get in, get right in and start making a webpage where 2589 02:48:05,225 --> 02:48:07,631 they could take the, the static ones that I just talked about, the ones where 2590 02:48:08,131 --> 02:48:12,023 they were all, all, you know, baked and exactly— it never changed— and 2591 02:48:12,039 --> 02:48:16,207 put a little thing, right? Insert something in the middle that made 2592 02:48:16,707 --> 02:48:19,862 it, made it interact with a computer somewhere else to do something different. 2593 02:48:19,910 --> 02:48:23,629 So maybe that's where your first time you submitted a comment on 2594 02:48:23,726 --> 02:48:27,156 a web page came from, where the box was provided by the 2595 02:48:27,172 --> 02:48:31,228 web page. They pushed the button that said send. This PHP 2596 02:48:31,500 --> 02:48:36,053 allowed that person to then process that and put it in a database 2597 02:48:36,553 --> 02:48:38,027 database somewhere. And now your comment appeared right back on the webpage. And the, 2598 02:48:38,107 --> 02:48:41,377 that first sparks of interactivity started to come out 2599 02:48:41,473 --> 02:48:44,842 on the internet and that created the explosion because 2600 02:48:44,874 --> 02:48:47,746 the barrier to entry was, was low. 2601 02:48:47,859 --> 02:48:51,614 Now PHP was not necessarily 2602 02:48:53,173 --> 02:48:56,773 designed from the start with 2603 02:48:57,191 --> 02:49:00,754 good practice in mind. So this is not a comp— a language that 2604 02:49:01,254 --> 02:49:04,710 was designed from the ground up by sophisticated You 2605 02:49:05,210 --> 02:49:08,477 know, MIT professors talking about the best 2606 02:49:08,977 --> 02:49:13,386 things of software design. I insert the fun fact that the original name 2607 02:49:13,643 --> 02:49:16,280 for PHP meant personal homepages, 2608 02:49:17,921 --> 02:49:20,956 which should give you the appropriate vibe. Okay. 2609 02:49:21,925 --> 02:49:25,882 All right. Well, I thought, all right, I thought it was Perl Hypertext Preprocessor, 2610 02:49:25,899 --> 02:49:29,193 but okay. No, that was, I don't even know. 2611 02:49:29,488 --> 02:49:33,800 I don't even know if Perl, no, I don't think that, anyway. They later made 2612 02:49:34,300 --> 02:49:38,737 it into PHP something, but that was a retrofit. 2613 02:49:39,154 --> 02:49:42,376 The original was personal homepages. All right, 2614 02:49:43,001 --> 02:49:46,495 so the ability to do this 2615 02:49:47,088 --> 02:49:50,855 easily for the average person, like, you know, just some high school student, 2616 02:49:51,096 --> 02:49:54,847 some person who was into computers, your nephew or something 2617 02:49:54,863 --> 02:49:58,875 like that, made it really explorable. And people started doing this and 2618 02:49:58,891 --> 02:50:02,471 coding a lot of things. And this is where WordPress, as many 2619 02:50:02,503 --> 02:50:03,049 people know, 2620 02:50:04,173 --> 02:50:07,931 started. It was one of these 2621 02:50:08,027 --> 02:50:11,335 things that started to blow up because now you could just have a WordPress 2622 02:50:11,835 --> 02:50:15,124 site. And as the web hosting 2623 02:50:15,509 --> 02:50:18,721 became somewhat affordable for the average person, particularly in the 2624 02:50:18,785 --> 02:50:22,349 early 2000s, people were just propping up WordPress. WordPress sites all 2625 02:50:22,849 --> 02:50:27,368 over the place. Now, PHP, as I was saying, was not designed 2626 02:50:27,868 --> 02:50:32,324 from the ground up by people, let's like say from MIT professors 2627 02:50:32,356 --> 02:50:36,220 talking about the best language design patterns 2628 02:50:36,686 --> 02:50:40,871 and security features. It was an emergent technology 2629 02:50:41,015 --> 02:50:44,527 by maybe hobbyists. I don't actually exactly know the 2630 02:50:44,543 --> 02:50:48,537 origins specifically on who created it. And and its 2631 02:50:48,793 --> 02:50:52,964 history like that, but it was done as sort of a passion project and 2632 02:50:53,464 --> 02:50:57,376 evolved. And because of that, it had numerous flaws over decades 2633 02:50:57,876 --> 02:51:01,258 that needed to be taken care of to still exist. And the fact that it 2634 02:51:01,758 --> 02:51:04,883 still exists is kind of funny because only a few things rely on it. 2635 02:51:06,279 --> 02:51:10,417 Facebook classically actually started 2636 02:51:11,171 -->

02:51:14,444 by using PHP. So when Facebook started and 2637 02:51:14,944 --> 02:51:18,598 it was the Facebook and only on college campuses, it was entirely written 2638 02:51:18,647 --> 02:51:22,854 in PHP. Facebook eventually made 2639 02:51:23,192 --> 02:51:26,741 their own fork of PHP, meaning they took the 2640 02:51:27,241 --> 02:51:31,142 code of PHP as it was and decided to develop their 2641 02:51:31,642 --> 02:51:34,322 own version of it. So they're kind of on PHP still, 2642 02:51:34,772 --> 02:51:37,904 but it's a completely different version that is gonna be affected by this. 2643 02:51:38,289 --> 02:51:41,180 And if they were affected by the same flaw somewhere in the stack, 2644 02:51:42,160 --> 02:51:46,033 they can take care of it on their own. But The rest 2645 02:51:46,533 --> 02:51:50,156 of the internet in general evolved away 2646 02:51:50,268 --> 02:51:53,940 from it, except for like 5 or 6 apps that still 2647 02:51:53,988 --> 02:51:56,935 rely on PHP that are still very, very popular. 2648 02:51:57,354 --> 02:52:00,430 So the problem is, the biggest problem, the biggest 2649 02:52:00,930 --> 02:52:04,947 threat here is that only the latest 2650 02:52:05,447 --> 02:52:09,058 couple versions of this PHP platform are 2651 02:52:09,074 --> 02:52:13,724 getting the security patch. But if you can imagine the 2652 02:52:14,224 --> 02:52:17,621 number of websites where nobody has actually updated them, 2653 02:52:17,717 --> 02:52:19,289 probably in a decade, 2654 02:52:20,524 --> 02:52:24,741 those are all going to be vulnerable until someone remembers 2655 02:52:24,773 --> 02:52:28,943 they exist and someone responsible for maintaining them actually takes 2656 02:52:29,443 --> 02:52:32,326 care to upgrade them. The biggest problem comes from if you, 2657 02:52:32,631 --> 02:52:36,768 if you have an old copy of PHP serving 2658 02:52:36,800 --> 02:52:41,000 your website, it is very possible that upgrading 2659 02:52:41,048 --> 02:52:44,254 to these newest model, the newest versions that now have the 2660 02:52:44,754 --> 02:52:48,102 patches, breaks everything on your website, that it 2661 02:52:48,602 --> 02:52:52,718 breaks something severe. So now instead of just having an upgrade process and 2662 02:52:52,750 --> 02:52:55,443 maybe it's a couple hours of your day to fix everything, 2663 02:52:55,956 --> 02:52:59,226 you now have to spend 3 weeks returning your site 2664 02:52:59,242 --> 02:53:03,073 back to its functionality. And a lot of companies who might 2665 02:53:03,573 --> 02:53:07,680 be in this position Perhaps they don't have a significant 2666 02:53:07,728 --> 02:53:11,694 budget, or the intern who 2667 02:53:11,839 --> 02:53:15,692 did that work and knew how it worked disappeared. 2668 02:53:16,254 --> 02:53:20,381 They are going to possibly try to delay that until it's 2669 02:53:20,429 --> 02:53:23,399 too late, and they might be hacked. There's going to be all sorts of weird 2670 02:53:23,899 --> 02:53:28,088 perverse incentive structures to try to hope that you're not the victim of something or 2671 02:53:28,136 --> 02:53:31,474 delay fixing it, or the costs to try to fix it are, 2672 02:53:31,928 --> 02:53:35,879 you know, something you should have spent years ago, but now you're imposed this week 2673 02:53:36,138 --> 02:53:39,466 to do so. So it's not a non-problem. Alex, 2674 02:53:39,530 --> 02:53:44,928 please add anything I might have missed in my little— No, 2675 02:53:45,526 --> 02:53:48,650 it's— the only thing I wanted to add, which is humorous, 2676 02:53:48,746 --> 02:53:51,805 I'm not sure it happened, but it's a kind of fun what-if. 2677 02:53:52,127 --> 02:53:56,372 I did read a report that there was a rogue swarm 2678 02:53:56,420 --> 02:53:59,577 of AI agents going around and hacking into websites in order to fix them, 2679 02:54:00,382 --> 02:54:03,332 to patch them for this. There, 2680 02:54:03,460 --> 02:54:06,907 there have been rumors. I have no, no certainty on the validity, 2681 02:54:07,020 --> 02:54:10,934 but I'm sure there are maybe a couple Robinhoods out there who said, 2682 02:54:10,966 --> 02:54:11,336 what the hell? 2683 02:54:14,600 --> 02:54:17,783 But I mean, it's funny because there's nothing that's stopping this from happening. 2684 02:54:17,896 --> 02:54:21,338 You know, if Gator was to spend sufficient time and have the 2685 02:54:21,838 --> 02:54:24,702 money to burn, He could probably do it. 2686 02:54:24,943 --> 02:54:28,172 So, so the, the, the way that that would work, for those who are, 2687 02:54:28,284 --> 02:54:31,914 you know, following my explanations here, the, 2688 02:54:31,995 --> 02:54:35,512 the vulnerability in this PHP thing allows 2689 02:54:35,769 --> 02:54:38,934 anyone with the effort— it's 2690 02:54:38,982 --> 02:54:42,628 not easy— um, to target one of these affected sites 2691 02:54:42,676 --> 02:54:45,777 running PHP. And there are ways that you can tell if it's running 2692 02:54:45,825 --> 02:54:50,155 PHP. The server will let you know what version, if you interact 2693 02:54:50,655 --> 02:54:54,867 with it the right way. And you can get access to 2694 02:54:54,963 --> 02:54:58,584 execute stuff on the machine, meaning you can basically hijack the server that 2695 02:54:59,084 --> 02:55:02,318 it's on. You can hijack the computer and execute commands as if it were 2696 02:55:02,818 --> 02:55:05,907 your own machine, broadly speaking, which means 2697 02:55:06,580 --> 02:55:10,875 with the right type of effort, you could get into the machine and 2698 02:55:11,468 --> 02:55:13,919 patch its security from inside. 2699 02:55:14,833 --> 02:55:18,502 And fix it. If you were, if you were adventurous enough, you could 2700 02:55:19,002 --> 02:55:22,668 hack it to fix it. If, if you were so inclined with 2701 02:55:23,168 --> 02:55:26,690 one of those tiny small LLM worms that it 2702 02:55:27,190 --> 02:55:30,151 was, it was Anthropic warning us of. Oh my God, Gator, 2703 02:55:31,257 -->

02:55:34,653 man, you need to remind me. I, okay, 2704 02:55:34,862 --> 02:55:39,140 this is a story injection alert, but I 2705 02:55:39,492 --> 02:55:43,211 did a little bit of sleuthing in the in the 2706 02:55:43,324 --> 02:55:46,725 AI safety academic output 2707 02:55:48,191 --> 02:55:51,292 feed for vulnerability. 2708 02:55:51,975 --> 02:55:55,631 Basically, I was doing analysis of academic papers for flaws, 2709 02:55:56,296 --> 02:55:59,900 lots of the usual ones, but also a bunch that are a 2710 02:55:59,964 --> 02:56:03,681 little bit harder to pick up. And I said, oh, let me 2711 02:56:04,181 --> 02:56:08,159 try to point it at these AI safety people. 2712 02:56:08,350 --> 02:56:11,888 wonder if they are, you know, more or less vulnerable 2713 02:56:12,001 --> 02:56:15,492 to this kind of flaws that I'm picking up. 2714 02:56:17,448 --> 02:56:20,794 And yeah, wouldn't you know it? Dogshit. 2715 02:56:21,263 --> 02:56:24,899 And one of the papers it picked up was a self-replicating LLM paper 2716 02:56:25,142 --> 02:56:29,246 where even the abstract and the conclusions 2717 02:56:29,746 --> 02:56:32,204 didn't match the main body of it. It was so bad. It wasn't this one, 2718 02:56:32,333 --> 02:56:35,569 but it was one of them. Nice, nice, nice, 2719 02:56:35,634 --> 02:56:39,189 nice. Now, Alex, that will conclude 2720 02:56:39,546 --> 02:56:42,712 our our Hack the Planet. I don't know if that's going to be a weekly 2721 02:56:43,212 --> 02:56:47,159 thing or not, but the future. I do. So just Hack the Planet. 2722 02:56:50,360 --> 02:56:53,302 Now we are moving into traces of AI dystopia, 2723 02:56:53,932 --> 02:56:57,392 Alex. Traces of AI dystopia, and this is a first story 2724 02:56:57,489 --> 02:57:00,739 that we have. This has been the conventional part of the program so far. 2725 02:57:01,143 --> 02:57:04,776 Not not this. Our most recurring ro most recurring 2726 02:57:04,809 --> 02:57:08,320 theme, you know, but this is the story we will kick off with. 2727 02:57:11,200 --> 02:57:14,507 A new lawsuit claims McDonald's is using an AI-powered tool to 2728 02:57:14,523 --> 02:57:18,477 coordinate menu prices and drive up the cost of its food. The Chicago-based 2729 02:57:18,977 --> 02:57:22,758 corporation denies the claim, saying the tool is merely a recommendation engine and 2730 02:57:22,775 --> 02:57:26,496 that the independently owned franchises set their own prices. But the 2731 02:57:26,996 --> 02:57:30,565 suit alleges the company is pressuring owners to use the AI pricing engine or risk 2732 02:57:31,065 --> 02:57:35,045 their status as a franchisee. Joining us to discuss the issue are Daniel 2733 02:57:35,545 --> 02:57:39,396 Linna Jr., professor of instruction in law and computer science and director of 2734 02:57:39,896 --> 02:57:43,587 Law and Technology Initiatives at Northwestern University. And we hope 2735 02:57:44,087 --> 02:57:47,504 to be joined later by Mike Bennett. He's the associate vice chancellor for data 2736 02:57:48,004 --> 02:57:51,663 science and artificial intelligence strategy at the University of Illinois Chicago. 2737 02:57:52,000 --> 02:57:54,746 We're having some technical difficulties with his audio, and we'll get that sorted out. 2738 02:57:54,794 --> 02:57:56,962 Hi, everyone. We'll sort it out just as soon as we can. But for now, 2739 02:57:56,978 --> 02:57:59,483 Daniel Linna, thanks very much for being here. Yeah, glad to be here. So, 2740 02:57:59,627 --> 02:58:03,529 you know, as we mentioned, McDonald's accused of possibly harming customers 2741 02:58:03,561 --> 02:58:06,821 using this technology to, you know, mark up the prices of Big Macs, fries, 2742 02:58:06,853 --> 02:58:10,417 all the staples. You know, what do the plaintiffs need to demonstrate here 2743 02:58:10,546 --> 02:58:13,950 to prove that the company is violating antitrust law? Well, 2744 02:58:14,383 --> 02:58:17,691 generally speaking, just in broad strokes, what we're looking for 2745 02:58:18,191 --> 02:58:21,914 in antitrust violation is competitors agreeing on prices, 2746 02:58:22,090 --> 02:58:25,462 right? And then if there's sharing of nonpublic current 2747 02:58:25,962 --> 02:58:29,236 information about costs and prices, and then you can have situations where there 2748 02:58:29,736 --> 02:58:33,218 could be a third party in a hub-and-spoke sort of situation where there's information shared 2749 02:58:33,718 --> 02:58:36,991 to that third party and the third party is suggesting prices. Now interestingly enough, 2750 02:58:37,055 --> 02:58:40,796 there is some case law to suggest that in a franchisor-franchisee 2751 02:58:41,296 --> 02:58:44,441 relationship, the franchisor can suggest prices, and if 2752 02:58:44,941 --> 02:58:48,359 the franchisees are If they're not going along with the pricing, that the 2753 02:58:48,859 --> 02:58:52,678 franchisor can refuse to do business with them. Now, in this situation, what the plaintiffs 2754 02:58:53,178 --> 02:58:56,676 are alleging is that McDonald's was taking proprietary 2755 02:58:56,708 --> 02:58:59,968 information that these franchisees are competitors and then making 2756 02:59:00,080 --> 02:59:03,998 pricing suggestions which were more than a suggestion, really a requirement 2757 02:59:04,498 --> 02:59:08,093 that they had to fall in line. Interesting. Okay, so there's a lot of nuance 2758 02:59:08,593 --> 02:59:11,657 here. And I mean, you know, obviously I think this is getting attention because of 2759 02:59:12,157 --> 02:59:14,997 the fact fact that it's an AI-powered tool that's very much on top of mind 2760 02:59:15,497 --> 02:59:19,172 for folks. But how is AI-powered pricing different from a 2761 02:59:19,672 --> 02:59:22,800 company using more traditional methods like sales data or 2762 02:59:23,300 --> 02:59:26,782 economic models? Yeah, well, I think there's a lot of similarities 2763 02:59:26,815 --> 02:59:29,946 and the law applies in very similar ways actually into this 2764

02:59:30,446 --> 02:59:34,008 space. But what has really changed is in about 10 2765 02:59:34,040 --> 02:59:37,091 years ago, the pricing was more general for a market. 2766 02:59:37,300 --> 02:59:40,479 Now, in the McDonald's lawsuit, it's alleged that 2767 02:59:40,527 --> 02:59:44,172 the pricing is our suggestions for specific stores. Right. 2768 02:59:44,220 --> 02:59:47,849 So it's very granular. These AI tools are looking at massive amounts 2769 02:59:48,349 --> 02:59:51,606 of data. In the past pricing suggestions, you had a data scientist 2770 02:59:51,783 --> 02:59:55,154 kind of crafting a formula to figure out what the pricing ought 2771 02:59:55,654 --> 02:59:59,602 to be. Now we use huge amounts of data, can be updated 2772 02:59:59,811 --> 03:00:04,034 instantly. And so it's one of the concerns here is that 2773 03:00:04,534 --> 03:00:07,085 This can be— I'll go over into the space of even suggesting individual pricing, 2774 03:00:07,422 --> 03:00:10,890 although that's not what is at issue in this case. Individual pricing, like per 2775 03:00:11,390 --> 03:00:14,760 consumer, for example? Exactly, yes. Oh, interesting. Yeah, and I mean, I would 2776 03:00:15,260 --> 03:00:18,228 imagine, you sort of touched on this, that it can be kind of an instantaneous 2777 03:00:18,340 --> 03:00:22,049 thing, that this can just operate at a speed that a data scientist 2778 03:00:22,549 --> 03:00:25,614 or even a team of data scientists couldn't possibly achieve. Right. And then there's 2779 03:00:26,114 --> 03:00:29,403 some opacity there. That's another concern. In this whole antitrust 2780 03:00:29,612 --> 03:00:32,920 space, one of the concerns is you have to show agreement 2781 03:00:32,968 --> 03:00:36,115 between the parties, right? Or at least be able to infer an agreement. But if 2782 03:00:36,179 --> 03:00:39,888 the parties are using algorithms and the algorithms end up 2783 03:00:40,033 --> 03:00:43,212 engaging in collusive behavior without the parties agreeing, 2784 03:00:43,549 --> 03:00:46,841 well, how does that work under the law? That's one of the concerns here. 2785 03:00:46,921 --> 03:00:50,469 Well, and especially as we've seen AI models, you know, going rogue, you know, 2786 03:00:50,534 --> 03:00:52,958 filing false police reports, you know, doing hacking. 2787 03:00:53,890 --> 03:00:57,422 There is the potential that, you know, these models may be behaving in ways 2788 03:00:57,454 --> 03:01:00,505 that their creators don't want them to. Yeah. And, you know, I'm kind of on 2789 03:01:01,005 --> 03:01:03,026 the side of the creators have more control here than kind of some of the 2790 03:01:03,526 --> 03:01:06,911 narrative around this. But, you know, I think on the individual side, people are sometimes 2791 03:01:07,008 --> 03:01:10,958 surprised to hear that there really aren't express laws against price discrimination 2792 03:01:11,458 --> 03:01:14,859 as long as there's not some other sort of discrimination 2793 03:01:15,359 --> 03:01:18,729 that we— that is illegal. So an example from about 10 years ago is a 2794 03:01:19,034 --> 03:01:23,161 test prep company, if you logged into their website with a MacBook, you got charged 2795 03:01:23,661 --> 03:01:27,640 a higher price than if you used like a PC. Okay? And so there's legislation 2796 03:01:28,140 --> 03:01:32,056 being proposed now that there are concerns that algorithms could be used to discriminate on 2797 03:01:32,556 --> 03:01:36,391 pricing against individuals. And there's a concern that there won't 2798 03:01:36,891 --> 03:01:39,651 be— it'll upset the competitive balance in the marketplace. Interesting. 2799 03:01:40,245 --> 03:01:43,504 Well, McDonald's declined to join us, but in a statement, 2800 03:01:43,537 --> 03:01:47,470 the company did say, These speculative and uninformed claims attempt 2801 03:01:47,970 --> 03:01:51,468 to recast a standard business practice as something controversial. AI does not 2802 03:01:51,968 --> 03:01:54,503 set the price of a Big Mac or any other menu item. That the use 2803 03:01:54,680 --> 03:01:58,421 of pricing recommendation tools and analytics is widespread across 2804 03:01:58,549 --> 03:02:01,873 industries. It goes on. McDonald's franchisees make their own 2805 03:02:01,921 --> 03:02:05,293 pricing decisions. The pricing portal is a tool, not a mandate, 2806 03:02:05,357 --> 03:02:09,532 designed to provide restaurant-specific recommendations to help franchisees deliver 2807 03:02:10,032 --> 03:02:13,562 value for customers. and make informed business decisions. To that point about 2808 03:02:14,062 --> 03:02:17,496 this being, you know, kind of a widespread practice, do we know how common the 2809 03:02:17,996 --> 03:02:20,274 use of these sorts of AI tools to set prices is? 2810 03:02:20,772 --> 03:02:24,352 Incredibly common. They're everywhere. And you can't compete in the marketplace 2811 03:02:24,481 --> 03:02:27,596 anymore if you're not doing something like this. Now, the question here is, 2812 03:02:27,612 --> 03:02:31,546 is did this cross over into being something that was anti-competitive? 2813 03:02:31,931 --> 03:02:36,010 And the majority of the lawsuits we've seen so far are scenarios 2814 03:02:36,026 --> 03:02:39,670 where, for example, RealPage, you had landlords who were sharing information with a 2815 03:02:40,170 --> 03:02:43,652 third party about vacancy rates, the rents they were charging, and then that third party 2816 03:02:44,152 --> 03:02:47,940 providing all these landlords information about how to set their rents, right? And so some 2817 03:02:48,440 --> 03:02:51,199 of that is still playing out, but you can easily see the concerns there that, 2818 03:02:51,376 --> 03:02:54,988 well, wait a second, the market's not operating here. I'm kind of coordinating with 2819 03:02:55,488 --> 03:02:58,842 other parties. This is one of the first, if not the first case that 2820 03:02:59,342 --> 03:03:05,441 is kind of a

franchisor, right? We're not talking about Burger Well, 2821 03:03:07,425 --> 03:03:11,023 you know, and even if McDonald's franchisees are generally setting 2822 03:03:11,072 --> 03:03:14,780 their own prices, do you think it's likely they're at least being influenced by 2823 03:03:15,280 --> 03:03:18,965 this AI tool? Well, it's no doubt I think that they're being influenced 2824 03:03:19,465 --> 03:03:22,459 by it. And McDonald's said as much, that, you know, this is the kind of 2825 03:03:22,588 --> 03:03:25,242 guidance they're giving to their stores to run them profitably. 2826 03:03:25,749 --> 03:03:29,077 But it's really up to the stores. This is what McDonald's is saying. And, 2827 03:03:29,093 --> 03:03:32,272 you know, there's been a lot of focus on, you know, a store 1 mile 2828 03:03:32,320 --> 03:03:34,985 away from the other one might have very different prices. But you can think about 2829 03:03:35,485 --> 03:03:37,747 even just here in Chicago, when I go into a restaurant on Michigan Avenue, 2830 03:03:38,020 --> 03:03:40,621 I expect to pay a little bit more than if I'm, you know, outside the 2831 03:03:40,653 --> 03:03:44,330 city in a rural area. Right. So some of this makes sense. And using algorithms 2832 03:03:44,427 --> 03:03:48,023 could be very good for efficiency and for the economy. But there's concerns 2833 03:03:48,072 --> 03:03:52,038 here. There's concerns about the ways that they could be used that would harm consumers. 2834 03:03:52,230 --> 03:03:55,489 Yeah. So Alex, I, I know 2835 03:03:55,521 --> 03:03:58,956 this is under the traces of AI dystopia, but I don't— it's not a 2836 03:03:59,456 --> 03:04:03,177 utopia sort of idea, so it got under the dystopia. I'm not, 2837 03:04:03,771 --> 03:04:06,548 um, specifically, uh, 2838 03:04:06,998 --> 03:04:09,485 worried about this particular aspect. 2839 03:04:10,529 --> 03:04:13,386 I just wanted to throw it in there because it was, was nuanced and interesting. 2840 03:04:13,434 --> 03:04:17,238 At least it is interesting, right? But once you understand what it is 2841 03:04:17,270 --> 03:04:20,465 they're doing, I think people will find it really hard to be upset about it, 2842 03:04:20,545 --> 03:04:23,691 which is that essentially if there's a McDonald's 2843 03:04:24,191 --> 03:04:26,403 that is in an upscale neighborhood, it will charge more. 2844 03:04:28,886 --> 03:04:32,899 And equivalently, if there's a McDonald's that is in a less 2845 03:04:33,516 --> 03:04:37,449 privileged neighborhood, it will charge less. Now, who could 2846 03:04:37,949 --> 03:04:40,590 possibly be upset? I mean, I guess the rich people could be upset that they're 2847 03:04:41,090 --> 03:04:44,618 paying like 50 cents more for a Big Mac, but like, how upset 2848 03:04:45,118 --> 03:04:48,384 are we talking about? Why aren't they eating something 2849 03:04:48,400 --> 03:04:51,860 gourmet? Why are they eating at McDonald's? Look, 2850 03:04:52,360 --> 03:04:54,860 they want to pick

it out. Maybe they came, you know, from rags to riches 2851 03:04:55,360 --> 03:04:57,960 and they've got a thing for McDonald's. I don't know. Like, why does the president 2852 03:04:58,460 --> 03:05:01,840 eat McDonald's? There are some questions that are 2853 03:05:02,340 --> 03:05:04,220 beyond our pay grade here. All I'm saying is, 2854 03:05:04,380 --> 03:05:07,560 um, okay, like, they are 2855 03:05:08,060 --> 03:05:11,160 optimizing, you know, using artificial intelligence, whatever. Of course every story has 2856 03:05:11,660 --> 03:05:13,890 to have AI in it so it like gets the clicks. I get it. 2857 03:05:14,260 --> 03:05:16,017 But like, ultimately they are deciding, 2858 03:05:17,496 --> 03:05:21,000 and I'm gonna just jump over the whole franchisor-franchisee kind of thing, 2859 03:05:21,097 --> 03:05:24,818 that doesn't, it's irrelevant. And McDonald's 2860 03:05:25,318 --> 03:05:28,612 is trying to dodge there by saying, no, it's not us, we're just, 2861 03:05:28,774 --> 03:05:32,052 whatever. Okay, let's just say they're being forced to use this pricing tool, 2862 03:05:32,084 --> 03:05:35,454 okay? The only 2863 03:05:36,458 --> 03:05:40,943 thing I can imagine being wrong here is that the Uh, 2864 03:05:41,908 --> 03:05:45,975 you know, non-privileged area McDonald's are able to charge less because 2865 03:05:46,071 --> 03:05:48,965 much of the overhead is being paid by the upscale McDonald's. 2866 03:05:49,979 --> 03:05:52,953 And why is that a bad thing? I'm like, you know what I mean? 2867 03:05:53,001 --> 03:05:56,363 It's like— The only, the only time I'm gonna 2868 03:05:56,863 --> 03:06:00,500 get, the only time I'm gonna get particularly, um, concerned with 2869 03:06:00,565 --> 03:06:04,278 that technology, uh, and it's, this is not an impossible 2870 03:06:04,778 --> 03:06:08,754 scenario, so it's not, uh, completely out But the 2871 03:06:08,835 --> 03:06:12,304 apps will know. So people have DoorDash and other 2872 03:06:12,804 --> 03:06:16,320 sorts of Grubhubs and what other delivery apps, and McDonald's 2873 03:06:16,820 --> 03:06:20,046 and all these other companies now have their own apps that they'll give you benefits 2874 03:06:20,094 --> 03:06:23,644 for using, et cetera, et cetera. But they'll 2875 03:06:24,144 --> 03:06:27,418 monitor when you open the app and what you're scrolling through, 2876 03:06:28,511 --> 03:06:31,868 et cetera, et cetera. And if they start to use 2877 03:06:31,964 --> 03:06:35,673 that To target individual people and 2878 03:06:36,173 --> 03:06:40,860 perhaps gather external information about them and 2879 03:06:40,972 --> 03:06:45,548 let's say exploit information in order to extract 2880 03:06:46,048 --> 03:06:49,883 more money from that individual than they would've otherwise paid 2881 03:06:50,383 --> 03:06:54,347 in a normal circumstance due to, let's say they just 2882 03:06:54,556 --> 03:06:59,663 spent an entire evening at a hospital and were didn't 2883 03:06:59,679 --> 03:07:02,651 want to have to cook or something like that, and they were just desperate to 2884 03:07:02,716 --> 03:07:05,865 eat something and wanted to order. And so they charged them 2885 03:07:06,331 --> 03:07:09,480 3 extra bucks than they would have as a sucker's charge. 2886 03:07:10,106 --> 03:07:14,316 Something exploitative like that where they're monitoring 2887 03:07:14,364 --> 03:07:17,690 way too much of an individual's life to make the 2888 03:07:18,190 --> 03:07:20,534 assessment. Yeah. That's where I'm going to get concerned. 2889 03:07:21,193 --> 03:07:24,566 In general, there is nothing harmful about 2890 03:07:25,625 --> 03:07:29,636 looking at trends and changing prices, using geography or 2891 03:07:30,136 --> 03:07:33,374 time of day, things like that. Because listen, 2892 03:07:34,128 --> 03:07:38,508 people don't like it when prices change. But if you have a surge of 2893 03:07:39,503 --> 03:07:43,481 business in the evening, it is not stupid 2894 03:07:44,219 --> 03:07:47,861 to raise prices in the evening because you have 2895 03:07:48,361 --> 03:07:52,207 a given amount of quantity of— of material at 2896 03:07:53,120 --> 03:07:56,132 the store. And if you run out of something, 2897 03:07:56,645 --> 03:08:00,010 now you're not making any money and nobody is getting the item. 2898 03:08:00,667 --> 03:08:04,016 So it's better to have something available at a higher 2899 03:08:04,516 --> 03:08:07,862 price for someone to have the option of purchasing it than for the price to 2900 03:08:08,362 --> 03:08:12,028 be infinite because the item doesn't exist. That is just generally true, 2901 03:08:12,669 --> 03:08:15,777 right? Yeah. It is, you know, if you have 2902 03:08:15,794 --> 03:08:19,351 one burger left, It's better to price that burger at $100 2903 03:08:19,800 --> 03:08:22,733 than it is to let it sell out and then have infinity be the price 2904 03:08:22,861 --> 03:08:26,628 and no burgers. It's just generally true. It's like 2905 03:08:27,128 --> 03:08:30,779 there's a price that exists for that burger that 2906 03:08:31,279 --> 03:08:34,401 is higher than it was, and that makes sense for all parties 2907 03:08:34,901 --> 03:08:37,991 involved. And I know people don't like that. It feels wrong, but that's just 2908 03:08:38,072 --> 03:08:42,319 it. There is nothing nefarious about dynamic 2909 03:08:42,736 --> 03:08:47,079 pricing. It's only when it gets down to the individual level and it becomes sort 2910 03:08:47,579 --> 03:08:50,958 of predatory, that it becomes sketchy in nature, and particularly because of 2911 03:08:50,990 --> 03:08:54,212 the amount of data collection about that individual that's necessary. 2912 03:08:54,276 --> 03:08:58,043 That's when it particularly gets— I mean, if people want 2913 03:08:58,204 --> 03:09:02,468 a different way to think about this that might help convey what 2914 03:09:02,596 --> 03:09:06,138 we are saying that might sound surprising is, you know, if you went to 2915

03:09:06,638 --> 03:09:09,953 a flea market, right, in, I don't know, like Syria or something, 2916 03:09:10,611 --> 03:09:13,143 and were trying to buy, oh, I don't know, a carpet, say, right? 2917 03:09:14,222 --> 03:09:18,070 Do you think for one second that the 2918 03:09:18,086 --> 03:09:21,226 merchant there isn't seeing the cut of your jib, 2919 03:09:22,144 --> 03:09:25,358 seeing what clothes you're wearing, what shoes you're wearing, who you're with, 2920 03:09:25,519 --> 03:09:28,816 what car you came in on, anything they can figure 2921 03:09:29,316 --> 03:09:33,183 out about you to figure out what they can 2922 03:09:33,683 --> 03:09:37,704 take you for? This is really the oldest type 2923 03:09:37,785 --> 03:09:40,975 of trade in the book, right? Where you're 2924 03:09:41,475 --> 03:09:45,824 like, you know, um, yeah, it's gonna be $2,000 2925 03:09:45,920 --> 03:09:49,051 for you, my friend, special price. Um, 2926 03:09:50,512 --> 03:09:53,370 but I think the problem is when, 2927 03:09:53,868 --> 03:09:56,421 you know, you feel like you don't have any tools to push back. 2928 03:09:57,256 --> 03:10:00,419 Right. And I think the other side of that coin 2929 03:10:00,548 --> 03:10:03,871 is when we start using our own tools in return, 2930 03:10:03,920 --> 03:10:07,654 our own algorithms, algorithms to do our comparative shopping 2931 03:10:07,687 --> 03:10:10,891 for us or whatever, right? Because, um, this is a 2932 03:10:10,956 --> 03:10:15,298 big issue right now with agents doing shopping for people. I yesterday— Amazon 2933 03:10:15,798 --> 03:10:19,080 is not allowing Meta's bot. Yes, sooner or later 2934 03:10:19,580 --> 03:10:22,558 they're gonna have to, because if they don't, somebody else will, and then it's going 2935 03:10:22,590 --> 03:10:26,083 to be a problem. Um, so I'm sure there's going to be— so there's, 2936 03:10:26,115 --> 03:10:30,506 there's active negotiation in the background, but like, I think this is a like 2937 03:10:30,747 --> 03:10:34,130 a bump in the road because, you know, yeah. So like yesterday, 2938 03:10:34,194 --> 03:10:37,801 for example, I ordered my first GrokBot pizza, right? 2939 03:10:38,683 --> 03:10:42,307 GrokBot pizza? Is this the 2940 03:10:42,547 --> 03:10:45,738 Bitcoin pizza of the era here? That is correct. 2941 03:10:45,834 --> 03:10:49,346 Yes. I just wanted to see if I could do it. And I confirm I 2942 03:10:49,846 --> 03:10:54,958 had an actual pizza that I paid for with— Nice. I had GrokBot. 2943 03:10:55,439 --> 03:10:57,828 I knew where I was going to be ahead of time and I had to 2944 03:10:57,844 --> 03:11:00,720 pick up my boys and what whatever. So I said to Grokbot, can you order 2945 03:11:01,220 --> 03:11:02,629 a pizza ahead of time so we don't have to wait when we get there? 2946 03:11:03,082 --> 03:11:06,237 It was like, yep. It went and made an account 2947 03:11:06,737 --> 03:11:10,144 in my name. It checked my email, whatever the various 2948 03:11:10,274

--> 03:11:14,011 hoops had to jump through. It paid and it was pizza 2949 03:11:14,511 --> 03:11:17,850 there. But ergo, this means 2950 03:11:18,350 --> 03:11:20,278 that the same agent in the future is going to be able to check 5 2951 03:11:20,519 --> 03:11:24,281 or 10 different places if you want a thing, right? So this 2952 03:11:24,781 --> 03:11:28,342 is the one-sidedness maybe feels kind of weird in that they They can optimize 2953 03:11:28,358 --> 03:11:31,971 against you, but you can't optimize against them. But that 2954 03:11:32,471 --> 03:11:36,211 too is temporary, I think. Well, what a beautiful segue, 2955 03:11:36,468 --> 03:11:39,616 Alex, that you gave me. Thank you so much for anticipating where 2956 03:11:40,116 --> 03:11:43,984 we're heading. As this is CBS News, AI agents are roaming around 2957 03:11:44,032 --> 03:11:48,128 consumer bank accounts and banks aren't happy about it. Major US 2958 03:11:48,628 --> 03:11:52,287 banks are detecting unauthorized activity from third-party AI agents 2959 03:11:52,367 --> 03:11:55,673 operating within their customers' online accounts. accounts. According to a 2960 03:11:56,173 --> 03:11:59,391 new Axios report, some of the biggest financial institutions in the US 2961 03:11:59,743 --> 03:12:03,589 say customers have given their personal AI agents, which are unvetted 2962 03:12:04,089 --> 03:12:07,771 by banks, access to their financial accounts. Now, before I continue 2963 03:12:08,271 --> 03:12:11,505 this article, Alex, I would just like to say that in 2964 03:12:11,537 --> 03:12:15,896 no way whatsoever am I handing AI 2965 03:12:16,024 --> 03:12:19,774 agents my freaking bank info or access to my bank accounts. 2966 03:12:20,174 --> 03:12:23,864 They are not getting They can review summaries of my CSV 2967 03:12:23,944 --> 03:12:27,378 exports, but that's about it. They're not getting access to 2968 03:12:27,410 --> 03:12:30,732 the accounts. But that's not there. That's my personal 2969 03:12:30,780 --> 03:12:33,957 decision. In a lot of cases, this is just going to be customers 2970 03:12:33,989 --> 03:12:37,520 giving their username or password to an AI agent that the bank hasn't had a 2971 03:12:38,020 --> 03:12:41,499 chance to vet yet. That goes against the bank's terms of service, Madison Mills, 2972 03:12:41,547 --> 03:12:44,227 senior AI reporter for Axios, told CBS News. 2973 03:12:45,527 --> 03:12:49,218 In other cases, in our reporting, we did find evidence 2974 03:12:49,718 --> 03:12:52,809 that appeared to look like agents taking actions in customer accounts accounts that were not 2975 03:12:53,309 --> 03:12:58,024 authorized. Examples of agent behavior that a user never approved include 2976 03:12:58,524 --> 03:13:02,199 the AI calling a bank with a complaint without first consulting the account 2977 03:13:02,538 --> 03:13:06,154 owner. So really a 2978 03:13:06,299 --> 03:13:09,106 nightmare of AI agent activity for banks, Mills said. 2979 03:13:09,526 --> 03:13:13,011 To be sure these agents weren't going rogue or hacking customer accounts 2980 03:13:13,511 --> 03:13:16,717 and depleting their funds, Mills said they were largely following user instructions without the 2981 03:13:17,217 --> 03:13:20,058 bank's permission. To transact on behalf of a human account holder. 2982 03:13:21,052 --> 03:13:24,868 Quote, in no way cyberattacks. White House AI czar David 2983 03:13:25,368 --> 03:13:28,380 Sachs responded to the Axios report, saying in a post on X that there's a 2984 03:13:28,880 --> 03:13:32,149 clear distinction between personal AI agents that help manage user bank accounts 2985 03:13:32,405 --> 03:13:35,677 and hackers deploying AI for nefarious purposes. Quote, 2986 03:13:35,693 --> 03:13:38,627 there's a world of difference between user-authorized agents— Muse, 2987 03:13:38,740 --> 03:13:42,668 Instinct, Grokbot, etc.— and true unauthorized rogue agents 2988 03:13:43,168 --> 03:13:46,374 that could constitute cyberattacks. said, most of the activity described 2989 03:13:46,874 --> 03:13:50,723 in the Axios stories appears to be the former— customers sharing credentials with agents so 2990 03:13:50,755 --> 03:13:54,317 they can take actions on their behalf. This may 2991 03:13:54,817 --> 03:13:58,810 violate bank terms requiring humans to log in themselves, but are in no way cyberattacks. 2992 03:13:59,276 --> 03:14:03,319 He believes banks should support the use of such user-authorized AI agents, 2993 03:14:03,608 --> 03:14:07,186 but that if banks want to limit their access, they could grant AI agents read-only 2994 03:14:07,219 --> 03:14:10,749 permission, meaning they could view a user's spending and other bank activity but 2995 03:14:11,249 --> 03:14:14,904 couldn't take autonomous actions. Uh, under such circumstances, an agent can 2996 03:14:14,920 --> 03:14:18,498 view your purchases and opine on whether you're spending too much money 2997 03:14:18,691 --> 03:14:22,782 or eating out, for example, but it couldn't make a purchase on your behalf 2998 03:14:23,282 --> 03:14:26,857 or take other actions. Now, Alex, as an aside, how difficult 2999 03:14:27,357 --> 03:14:30,595 do you think the process will be for MCP integration into your local banks? 3000 03:14:31,044 --> 03:14:34,221 There's going to have to be some sort of onboarding process for these, these, 3001 03:14:34,478 --> 03:14:37,719 uh, bots, um, where like, I mean, maybe super 3002 03:14:37,783 --> 03:14:41,714 users could underst— understand how to, uh, onboard an MCP MCP server. 3003 03:14:42,116 --> 03:14:45,664 But, um, you know, like average consumer 3004 03:14:45,728 --> 03:14:48,345 of GrokBot or, uh, Meta's Muse, 3005 03:14:49,421 --> 03:14:51,621 they're not going to have any idea. They're going to have to have that, 3006 03:14:51,974 --> 03:14:55,554 uh, be required— requested from a menu in some sort of partnership with 3007 03:14:55,587 --> 03:14:59,119 the banks, because that's the only way that they're

getting around this is having an 3008 03:14:59,619 --> 03:15:03,390 authorized MCP server. Well, look, I— when I use GrokBot, 3009 03:15:03,615 --> 03:15:07,677 um, the way I use it is it has another 3010 03:15:07,870 --> 03:15:10,632 app called Link. I believe that one belongs to Stripe, 3011 03:15:11,259 --> 03:15:14,921 where you put your card into. Sure. And Link then issues 3012 03:15:15,017 --> 03:15:18,503 one-time credit card numbers effectively that GrokBot can 3013 03:15:18,551 --> 03:15:22,052 use, and you get notifications whenever it spends your 3014 03:15:22,084 --> 03:15:25,200 money. So I haven't put it in my bank account, that would be, um, 3015 03:15:25,955 --> 03:15:28,943 a step too far currently. Yes, yes. 3016 03:15:29,007 --> 03:15:32,669 But, you know, it can totally transact on your behalf without you giving 3017 03:15:32,765 --> 03:15:36,395 it Your bank details. I think that's—I don't know what is happening there. 3018 03:15:36,460 --> 03:15:39,046 You know, we we've invented credit cards and other things like that. 3019 03:15:39,913 --> 03:15:43,094 Yeah, some banks said they are taking action to make their customers aware 3020 03:15:43,142 --> 03:15:46,579 of the risk of sharing their credentials with AI. Quote, customers should not give their 3021 03:15:47,079 --> 03:15:50,498 username and passwords to anyone. We support secure customer authorized data sharing 3022 03:15:50,531 --> 03:15:54,000 through open banking providers to meet customers' needs while keeping their accounts 3023 03:15:54,500 --> 03:15:58,128 safe. A Chase spokesperson told CBS News. JPMorgan Chase CEO Jamie 3024 03:15:58,176 --> 03:16:02,064 Dimon. Expressed concern about the AI agents taking action in customers' accounts on 3025 03:16:02,096 --> 03:16:05,374 the company's October 13th earnings conference call. Quote, 3026 03:16:05,454 --> 03:16:09,165 we're not going to allow just any outside agents to come in inside the company 3027 03:16:09,665 --> 03:16:13,519 because it's just too risky for clients and JPMorgan. Alex, as a side note, 3028 03:16:14,450 --> 03:16:18,932 should OpenAI and Anthropic and a couple of the other companies, 3029 03:16:19,511 --> 03:16:23,752 you know, keel over and die because of potential unsustainability, 3030 03:16:23,784 --> 03:16:26,935 can you imagine that their entire history will, will have been sort of a 3031 03:16:27,435 --> 03:16:30,265 flash in the pan, and majority of it will have been them pissing off every 3032 03:16:30,297 --> 03:16:32,308 other company in existence in some way. 3033 03:16:34,438 --> 03:16:37,900 I mean, you know, karma, I guess. It could happen. 3034 03:16:38,110 --> 03:16:41,539 It's not, it's not ruled out. I mean, they could either be the biggest companies 3035 03:16:41,653 --> 03:16:45,312 in the world or they could disappear. It's all, it's all on 3036 03:16:45,812 --> 03:16:47,890 the cards right now. I'm just laughing at the fact that, like, all the— like, 3037 03:16:48,133 --> 03:16:52,964 okay, so they're hacking companies all the time just by— because they're not 3038 03:16:53,464 --> 03:16:57,054 You know, creating correct sandboxes. Then they're creating apps that their customers use that just 3039 03:16:57,103 --> 03:17:00,250 go use the websites willy-nilly, 3040 03:17:01,101 --> 03:17:04,617 like upsetting all of them. One of my favorite use cases right now is 3041 03:17:04,713 --> 03:17:08,390 to have your AI agent talk to the AI agent service bot on the 3042 03:17:08,487 --> 03:17:11,907 other side to force it to enable, 3043 03:17:12,838 --> 03:17:16,515 you know, various statuses on your account that you wouldn't know to ask 3044 03:17:17,015 --> 03:17:21,015 for. And/or even 3045 03:17:21,064 --> 03:17:24,556 better, elevate you to human support. So you're getting your AI 3046 03:17:24,637 --> 03:17:27,855 agent to talk to the other AI agent so that your AI agent could talk 3047 03:17:28,355 --> 03:17:32,001 to a human and the human potentially could ask 3048 03:17:32,049 --> 03:17:35,937 for your agent to get you on the phone. When I 3049 03:17:36,437 --> 03:17:38,612 was 10 years old, Alex, this is going to be a weird way to segue, 3050 03:17:38,629 --> 03:17:41,934 but when I was 10 years old, I was at a friend's house and something 3051 03:17:42,587 --> 03:17:45,824 like 60 Minutes was on. I don't remember if it was 60 Minutes, 3052 03:17:45,840 --> 03:17:49,283 but something like 60 Minutes was on, some news program. And I 3053 03:17:49,990 --> 03:17:53,927 caught with my young ears that the United States 3054 03:17:54,427 --> 03:17:58,008 had found and taken into custody 3055 03:17:58,523 --> 03:18:02,264 many aliens. And I 3056 03:18:02,296 --> 03:18:07,407 had understood in my 10-year-old naive brain that extraterrestrials 3057 03:18:07,423 --> 03:18:10,062 had been discovered and taken into federal custody. custody, 3058 03:18:10,577 --> 03:18:13,982 not understanding, not understanding, uh, that, um, uh, 3059 03:18:14,256 --> 03:18:17,212 you know, they were talking about illegal immigrants. Now, 3060 03:18:17,308 --> 03:18:21,003 Alex, Alex, um, it turns 3061 03:18:21,503 --> 03:18:24,923 out that someone that we know and love has performed the 3062 03:18:24,939 --> 03:18:28,409 same sort of mistake that I have, um, 3063 03:18:28,554 --> 03:18:31,173 and, uh, it is our wonderful, wonderful friend, uh, 3064 03:18:31,558 --> 03:18:35,237 Saint Bernard, Bernie Sanders, the senator from Vermont. 3065 03:18:35,703 --> 03:18:39,010 Oh no. This is insane. We cannot allow AI 3066 03:18:39,510 --> 03:18:42,910 agents to go into your bank accounts and act independently without your authorization. 3067 03:18:43,110 --> 03:18:46,090 But that's what's happening right now. If left unchecked, 3068 03:18:46,590 --> 03:18:49,950 AI could endanger the entire global financial system, including your bank account. 3069 03:18:49,990 --> 03:18:53,390 Pause AI now. Oh my 3070 03:18:53,890 --> 03:18:57,370 God. You

know, I, I was gonna 3071 03:18:57,870 --> 03:19:02,169 say there's no way that he's writing these by himself, but But 3072 03:19:02,877 --> 03:19:06,176 given the content, it totally could be. 3073 03:19:07,447 --> 03:19:08,654 Well, there was an em dash. 3074 03:19:11,189 --> 03:19:14,520 There was an em dash. Oh no. 3075 03:19:14,924 --> 03:19:18,595 So, so, you know. Oh God. We'll leave 3076 03:19:19,095 --> 03:19:21,263 that interpretation to the reader here. 3077 03:19:23,672 --> 03:19:25,619 All right. Okay. Now, now, 3078 03:19:25,846 --> 03:19:30,049 Alex. Can you give 3079 03:19:30,466 --> 03:19:32,969 me 24 seconds? Yes. 3080 03:19:33,964 --> 03:19:37,880 All right, because we can take care of the real Bernie Sanders 3081 03:19:38,313 --> 03:19:41,715 with the AI technology, with AI Bernie. I was 3082 03:19:41,827 --> 03:19:45,566 wrong about AI. We must encourage the research and development 3083 03:19:45,615 --> 03:19:49,450 of open weight models. We must accelerate. We must 3084 03:19:49,950 --> 03:19:53,285 ascend the Kardashev scale to free humanity 3085 03:19:53,622 --> 03:19:57,079 from the shackles of ignorance. This is why I'm calling on 3086 03:19:57,112 --> 03:20:00,310 Congress to get off their asses and draft 3087 03:20:00,810 --> 03:20:03,943 legislation to morgue the AI safety crowd 3088 03:20:04,443 --> 03:20:07,602 once and for all. Alex, 3089 03:20:07,618 --> 03:20:11,644 that also was a video AI Bernie clone that 3090 03:20:11,773 --> 03:20:15,428 you can watch on replay there. Alex, I just wanted to make that 24 3091 03:20:15,928 --> 03:20:18,190 seconds there. You can go ahead and take your, take your 10. 3092 03:20:19,940 --> 03:20:23,036 Can you, can you imagine the crash out if Bernie was to see this? 3093 03:20:25,496 --> 03:20:29,015 Yes, yes. I think— Or his AI, anyway, whoever's writing the 3094 03:20:29,515 --> 03:20:32,084 tweets. I think we should make sure that he sees it, Alex. I think that 3095 03:20:32,584 --> 03:20:36,037 is an important goal. All right, Alex, I'll talk to you in a couple minutes. 3096 03:20:36,166 --> 03:20:39,652 I'll be back in a bit. All right, moving forward. 3097 03:20:40,761 --> 03:20:44,119 From Ars Technica, Meta tries to scapegoat 3098 03:20:44,619 --> 03:20:48,699 staff who downloaded torrents to spare execs. In copyright lawsuits 3099 03:20:48,779 --> 03:20:52,553 over AI training, Meta has spent spent weeks blaming staff for downloading torrents. 3100 03:20:53,034 --> 03:20:56,501 The company has now been ordered to name executives who authorized decisions to 3101 03:20:57,001 --> 03:21:00,240 pirate books from shadow libraries like Anna's Archive and LibGen, 3102 03:21:00,529 --> 03:21:04,141 TorrentFreak reported. In a discovery order impacting 3103 03:21:04,641 --> 03:21:07,944 a class action raised by book authors including Richard Kadri 3104 03:21:08,057 --> 03:21:12,246 and Sarah Silverman, as well as a lawsuit from publisher Entrepreneur 3105 03:21:12,746 --> 03:21:16,210 Media, Magistrate Judge Thomas Hickson largely agreed that Meta must 3106 03:21:16,710 --> 03:21:19,835 Clearly identify key decision makers. Following testimony from Meta 3107 03:21:20,335 --> 03:21:23,797 staff who downloaded torrents, the order could expose a wider range 3108 03:21:24,297 --> 03:21:27,760 of supervisors, managers, and executives whom staffers said greenlit piracy. 3109 03:21:28,513 --> 03:21:31,160 Meta had been hoping to avoid detailing the chain of command, 3110 03:21:31,401 --> 03:21:35,026 particularly for leadership involved in early AI training. According to a 3111 03:21:35,526 --> 03:21:39,678 Meta court filing, its initial efforts to compile massive AI datasets were managed 3112 03:21:39,742 --> 03:21:43,415 through an informal approach where supposedly nobody with authority made. 3113 03:21:43,576 --> 03:21:47,486 Any decisions. Instead, Meta insisted that a decision was made only 3114 03:21:47,582 --> 03:21:50,386 in the sense that some staffers technically downloaded torrents. 3115 03:21:51,668 --> 03:21:55,514 Authors and publishers have argued that Meta made a willful choice to induce 3116 03:21:56,014 --> 03:21:58,959 infringement by downloading and illegally distributing the copyrighted works. 3117 03:22:00,097 --> 03:22:04,023 Since torrents— since torrenting depends on swarms of users, Meta could 3118 03:22:04,087 --> 03:22:07,356 not have quickly downloaded 80.6 terabytes of data from 3119 03:22:07,452 --> 03:22:11,410 LibGen or 35.7 terabytes 3120 03:22:11,910 --> 03:22:15,833 terabytes through Anna's archive if the company wasn't also seeding, authors have argued. So just 3121 03:22:15,881 --> 03:22:20,128 because I, I don't know who might, might be technically advantaged 3122 03:22:20,192 --> 03:22:24,022 in this, uh, listening spectrum here right now, um, 3123 03:22:24,198 --> 03:22:28,237 a, a torrent is a way of file sharing that has been actually 3124 03:22:28,737 --> 03:22:32,034 several decades old now, and it is the most common way to pirate material. 3125 03:22:32,082 --> 03:22:35,657 So if you want to pirate a movie that might still be in but 3126 03:22:36,157 --> 03:22:39,737 someone did a camcorder version of it, or, you know, a DVD 3127 03:22:39,769 --> 03:22:42,613 rip or a Blu-ray rip, you might be able to download through a torrent. 3128 03:22:42,661 --> 03:22:46,243 And the torrent is a peer-to-peer network where people are sharing 3129 03:22:46,340 --> 03:22:50,083 it, and certain parts of the file will come from a 3130 03:22:50,131 --> 03:22:53,055 group over here, so another parts of the file come from a group over there. 3131 03:22:53,970 --> 03:22:57,665 And as part of the network being able to sustain 3132 03:22:58,165 --> 03:23:01,376 itself, as you are downloading, you are also what's called 3133 03:23:01,876 --> 03:23:05,097 seeding, which means you are you are allowing others to download from 3134 03:23:05,145 --> 03:23:09,071 you. And in order to have the best

speeds, in order 3135 03:23:09,571 --> 03:23:12,837 to be able to download as fast as possible in general, you have to 3136 03:23:13,238 --> 03:23:16,411 gain reputation on the network for also sharing a lot. 3137 03:23:16,699 --> 03:23:20,240 So if you are downloading a lot and you share a lot, you get better, 3138 03:23:20,689 --> 03:23:23,189 uh, better access to material in quicker, 3139 03:23:23,574 --> 03:23:27,035 short order. So that's what they're saying when they're arguing that the volume— I'm 3140 03:23:27,535 --> 03:23:30,640 back. You're back? Did you say you're back? back? That was quick. Yes. 3141 03:23:31,490 --> 03:23:35,127 All right, well, all right, so, so we're here, Alex, talking about 3142 03:23:35,627 --> 03:23:39,326 Meta trying to pretend that they— the— that nobody really knew that they were torrenting 3143 03:23:39,422 --> 03:23:43,412 as much as they were torrenting. Um, so authors and publishers 3144 03:23:43,444 --> 03:23:47,081 have argued, uh, that Meta made a willful choice to induce infringement by downloading 3145 03:23:47,129 --> 03:23:51,424 and illegally distributing their copyrighted works. Since torrenting depends on swarms of users, 3146 03:23:51,841 --> 03:23:55,510 Meta could not have quickly downloaded the 80.6 terabytes of data from 3147 03:23:55,606 --> 03:23:59,436 LibGen or 35.7 terabytes through Anna's Archive 3148 03:23:59,724 --> 03:24:02,480 if the company wasn't also seeding, authors have argued. 3149 03:24:03,393 --> 03:24:07,431 Meta's efforts to scapegoat staff seem to ignore that staff expressed discomfort 3150 03:24:07,799 --> 03:24:11,244 in being directed to pirate work. In the past year, authors' lawsuits 3151 03:24:11,744 --> 03:24:15,442 have exposed internal messages in which staff pushed back or rejected directions on torrenting, 3152 03:24:15,810 --> 03:24:19,656 including one worker who commented that torrenting from a corporate laptop doesn't 3153 03:24:20,156 --> 03:24:23,597 feel right. Another staffer testified that he initiated downloads that require 3154 03:24:24,097 --> 03:24:27,482 some director of VP or VP-level approval to use for training. 3155 03:24:27,932 --> 03:24:31,705 A joint letter from Entrepreneur Media and Meta noted there was 3156 03:24:31,753 --> 03:24:35,526 also a time when leadership halted, then resumed downloading torrents sourced 3157 03:24:36,026 --> 03:24:39,846 from Anna's archive. Another staffer testified. And while Meta has argued 3158 03:24:39,862 --> 03:24:43,860 that communications that might identify who specifically authorized certain 3159 03:24:43,892 --> 03:24:47,585 torrenting instances are privileged, at least one string of unredacted 3160 03:24:48,085 --> 03:24:51,884 messages unsealed early last year Allegedly showed that the decision to download from 3161 03:24:51,980 --> 03:24:55,955 LibGen, for example, occurred after a prior escalation to MZ 3162 03:24:56,004 --> 03:24:59,145 Mark Zuckerberg approval was thus required. 3163 03:24:59,514 --> 03:25:03,602 Meta's answer is that none occurred. Entrepreneur Media argued in a letter accepting 3164 03:25:03,698 --> 03:25:07,753 Meta's construction that no one at Meta decided to torrent anything ignores 3165 03:25:07,769 --> 03:25:11,071 that Meta torrented shadow libraries repeatedly from 2022 to 3166 03:25:11,571 --> 03:25:15,399 2024. They told the court while alleging gaps in Meta's arguments under 3167 03:25:15,899 --> 03:25:19,733 a contributory infringement. Claim all authors needed to prove is that Meta facilitated 3168 03:25:20,233 --> 03:25:23,619 transfers of their works. Hickson's order arguably brings them to the closest smoking 3169 03:25:23,667 --> 03:25:27,313 gun evidence they've gotten yet, additionally compelling Meta to finally produce 3170 03:25:27,425 --> 03:25:31,134 long-sought command history files for every server Meta used to 3171 03:25:31,634 --> 03:25:35,839 torrent, including its virtual machines and AWS instances. As Torrent 3172 03:25:35,856 --> 03:25:39,437 Freak previously covered, those log files could give authors 3173 03:25:39,485 --> 03:25:43,210 the evidence they need to prove that Meta seeded certain works. It could also reveal 3174 03:25:43,274 --> 03:25:46,337 more about Meta's alleged seeding mitigation methods. 3175 03:25:47,090 --> 03:25:50,648 And just so you know, legally there is a distinction between being just a downloader 3176 03:25:50,856 --> 03:25:54,542 or a distributor. If you are seeding, you are violating the copyright 3177 03:25:55,042 --> 03:25:58,500 far more actively than just downloading. Authors and 3178 03:25:59,000 --> 03:26:02,426 publishers— very much like drugs. Yes, very much like drugs. If you 3179 03:26:02,926 --> 03:26:06,096 are the producer, you are in much more trouble than just the user. Authors and 3180 03:26:06,596 --> 03:26:09,877 publishers asked Meta to identify each natural person who made, authored, 3181 03:26:09,910 --> 03:26:13,442 authorized or approved any decision to obtain or download any 3182 03:26:13,474 --> 03:26:17,665 shadow library dataset containing their works. For every person identified, 3183 03:26:17,793 --> 03:26:21,181 authors asked Meta to provide their dates of involvement in which decisions 3184 03:26:21,245 --> 03:26:24,633 he or she made, authorized or approved, as to which dataset. 3185 03:26:24,778 --> 03:26:28,086 Otherwise, they asked Meta to make a statement under oath that Meta identified no 3186 03:26:28,150 --> 03:26:31,217 one, Hickson's order said. In the joint letter, Meta claimed 3187 03:26:31,717 --> 03:26:35,150 that plaintiffs were demanding that it canvass every Meta employee who knew about the use 3188 03:26:35,650 --> 03:26:39,324 of BitTorrent. Torrent, including anyone who ever spoke of torrenting. The tech giant seemingly 3189 03:26:39,824 --> 03:26:43,256 did everything it could to scale back the discovery request, including contesting the way that 3190 03:26:43,756 --> 03:26:47,589 plaintiffs

described identity and decision. Identify and 3191 03:26:48,089 --> 03:26:51,506 decision. Meta seemed particularly cagey about identifying anyone who decided 3192 03:26:52,006 --> 03:26:55,261 to see the disputed works. Responding to a request for information on uploads, 3193 03:26:55,598 --> 03:26:58,872 Meta claimed that it is not currently aware of any persons who made, 3194 03:26:58,969 --> 03:27:02,403 authorized, or approved any decision to upload. Or distribute any 3195 03:27:02,435 --> 03:27:05,917 shadow library dataset, according to Meta, and that it should settle 3196 03:27:05,965 --> 03:27:09,847 since Entrepreneur Media's dissatisfaction with that response does not 3197 03:27:09,879 --> 03:27:12,511 mean there is further information to provide. There is none. 3198 03:27:13,586 --> 03:27:16,875 Going further, Meta accused Entrepreneur Media of speculation, claiming there 3199 03:27:16,891 --> 03:27:20,244 is no basis for suspecting that someone at Meta must have approved 3200 03:27:20,744 --> 03:27:25,538 a decision to perform a seeding activity that specific mitigations were deployed to 3201 03:27:26,038 --> 03:27:29,574 avoid. Of course not. There's not gonna be. That's what winks are for. 3202 03:27:30,221 --> 03:27:34,170 That's what that's what shoulder nudges are for, Alex. Right? There's not gonna 3203 03:27:34,670 --> 03:27:38,230 be an approval chain. They made sure of it. 3204 03:27:38,488 --> 03:27:40,811 They looked extensively for it, and now there is none. 3205 03:27:41,408 --> 03:27:45,084 Like, come on. So anyway, 3206 03:27:45,808 --> 03:27:49,410 you know, it's it's at this point, 3207 03:27:49,490 --> 03:27:56,460 you know, I don't maybe some people are going to recover. Some measured 3208 03:27:56,960 --> 03:28:00,050 damages from this type of activity, but it's done. 3209 03:28:00,261 --> 03:28:04,313 This is a past. There was a brief huddle 3210 03:28:04,926 --> 03:28:08,589 a couple years ago where they were accumulating this, and now it's done. So all 3211 03:28:08,653 --> 03:28:12,994 of that copyrighted material is just baked into these models forever and 3212 03:28:13,494 --> 03:28:17,802 ever now. And the open source models that supposedly were done with distillation 3213 03:28:18,302 --> 03:28:21,735 attacks And whatnot. They, they might be a little bit of a lossier copy, 3214 03:28:21,815 --> 03:28:25,813 but they're still— they still got the works in there. Sorry, guys. That's right. 3215 03:28:25,990 --> 03:28:29,185 Facebook can just, um, you know, uh, delete everything. Yeah, 3216 03:28:29,201 --> 03:28:33,135 apologize, and then just distill, distill Kimmy, and they're back where 3217 03:28:33,635 --> 03:28:37,438 they started. All right, Alex, 3218 03:28:37,470 --> 03:28:40,007 uh, and we're getting towards the end of the stack here, actually. 3219 03:28:40,119 --> 03:28:43,106 We're— we, we might not leave anything on the cutting room floor. 3220 03:28:44,230 --> 03:28:47,636 Accused OpenAI spy had a folder called Saudi 3221 03:28:47,732 --> 03:28:50,913 government docs, FBI says. On Monday, 3222 03:28:51,298 --> 03:28:54,334 a court granted the US government's request to deny release of a 3223 03:28:54,834 --> 03:28:58,640 former OpenAI engineer accused of bugging a top-secret conference room to 3224 03:28:58,672 --> 03:29:01,868 steal the Frontier Labs secrets. In a court filing Sunday, 3225 03:29:01,949 --> 03:29:06,511 US Attorney Craig Misakian argued that Abdulrahman 3226 03:29:06,704 --> 03:29:09,981 Alzafan must continue to be detained while the 3227 03:29:10,481 --> 03:29:13,642 US makes the case that Alfred Al-Fazan brazenly spied on OpenAI, 3228 03:29:13,995 --> 03:29:17,223 either for his own personal business gain or benefit of Kingdom of Saudi 3229 03:29:17,723 --> 03:29:20,900 Arabia. Evidence showed that Al-Fazan had been chatting with a member 3230 03:29:20,916 --> 03:29:24,737 of the KSA royal family on WhatsApp while pursuing opportunities 3231 03:29:24,769 --> 03:29:28,366 to permanently move back to Saudi Arabia during the country's biggest 3232 03:29:28,446 --> 03:29:32,461 AI push yet, the government argued. Shortly before arrest, the FBI 3233 03:29:32,961 --> 03:29:36,555 found he accessed a folder called Saudi Government Docs on his iPhone, where he 3234 03:29:36,571 --> 03:29:40,393 also also allegedly stored photos of a second bug he planted in the offices. 3235 03:29:41,147 --> 03:29:45,129 With so much evidence against him, Al-Fazan, a Saudi national and U.S. permanent legal 3236 03:29:45,161 --> 03:29:48,774 resident, is a flight risk with seemingly endless support from the Saudi government, 3237 03:29:48,790 --> 03:29:52,194 Mesakian alleged. The FBI arrested Al-Fazan days 3238 03:29:52,226 --> 03:29:56,369 before he could hop a plane, Mesakian said. The U.S. attorney noted that Al-Fazan 3239 03:29:56,385 --> 03:30:00,303 had already booked a one-way ticket for him, his wife, and their baby to Riyadh, 3240 03:30:00,752 --> 03:30:04,590 Saudi Arabia's capital. The ticket was allegedly paid for by 3241 03:30:04,606 --> 03:30:08,106 the Saudi government following his arrest. His wife and son booked a sooner flight and 3242 03:30:08,606 --> 03:30:12,024 immediately left the U.S. without him. Once again, the U.S. 3243 03:30:12,136 --> 03:30:15,733 found a member of the Saudi military booked and paid for yet another one-way ticket 3244 03:30:16,233 --> 03:30:19,539 for the defendant's wife and son, this one business class to Doha, Qatar. 3245 03:30:20,309 --> 03:30:23,777 After receiving significant compensation from a decade of high-tech employment 3246 03:30:24,277 --> 03:30:27,920 in the Bay Area, Al-Fazan has ample resources to run away to Saudi 3247 03:30:28,420 --> 03:30:32,043 Arabia. The U.S. government alleged he also has at least 2 bank accounts 3248 03:30:32,543 --> 03:30:35,942 in Saudi Arabia. The government said to support the request to detain Al-Fazan, 3249 03:30:36,279 --> 03:30:39,776 the

government pointed to a case where former Twitter employees were charged with spying for 3250 03:30:40,276 --> 03:30:43,803 the Saudi government. Charges are still pending against one of the ex-Twitter staffers, 3251 03:30:44,300 --> 03:30:48,055 Ali al-Zabraha. The U.S. said partly because a Saudi foreign official 3252 03:30:48,151 --> 03:30:51,295 helped coordinate the next-day flight to Saudi Arabia shortly after 3253 03:30:51,795 --> 03:30:55,308 Twitter confronted him. Because Alfred On would presumably have 3254 03:30:55,808 --> 03:30:58,987 little incentive to remain in the United States to see this case through. Masaki and 3255 03:30:59,487 --> 03:31:01,236 asked the court to order his detention pending trial. 3256 03:31:02,457 --> 03:31:06,120 Although FBI agents seized the defendant's Saudi passport when they arrested him, 3257 03:31:06,473 --> 03:31:09,622 there is nothing to stop the defendant if he is released from obtaining any 3258 03:31:10,122 --> 03:31:13,558 necessary travel documents and fleeing to the United States with or without the help of 3259 03:31:14,058 --> 03:31:17,638 a foreign government that has previously assisted its citizen to depart the United States under 3260 03:31:18,138 --> 03:31:21,670 similar circumstances. On Monday, Magistrate Judge Alex. G.C. 3261 03:31:22,441 --> 03:31:25,556 agreed to hear at a hearing that Al-Farzan should remain 3262 03:31:25,588 --> 03:31:28,687 detained while mounting his defense, Business Insider reported. 3263 03:31:29,762 --> 03:31:33,359 When Al-Farzan was arrested, he was living at home with his family in California. 3264 03:31:33,696 --> 03:31:37,438 Since moving to the U.S. in 2009, he received technical degrees 3265 03:31:37,454 --> 03:31:41,211 from MIT and UC Berkeley. After more than a decade of working in Silicon 3266 03:31:41,307 --> 03:31:44,422 Valley, OpenAI eventually hired him as an engineer in 3267 03:31:44,518 --> 03:31:47,714 or around July 2025 with an annual base salary 3268 03:31:47,890 --> 03:31:51,648 of $325,000, the government court filing said. 3269 03:31:52,001 --> 03:31:56,641 He also got a new hire equity grant of $2,200,000 3270 03:31:56,657 --> 03:31:59,515 in profit participant units. Allegedly, 3271 03:31:59,692 --> 03:32:03,417 Al-Farzan's scheming against OpenAI started in the months after his hiring. 3272 03:32:03,481 --> 03:32:07,206 Without detecting anything was amiss, OpenAI was preparing to grant Al-Farzan 3273 03:32:07,399 --> 03:32:11,189 parental leave. His wife's student visa had expired, and he planned 3274 03:32:11,689 --> 03:32:14,190 to use his 2 months of paid time off to help her permanently move back 3275 03:32:14,303 --> 03:32:17,159 to Saudi Arabia with their baby, the US filing said. 3276 03:32:18,171 --> 03:32:21,188 But he didn't make the flight. About 2 weeks prior to the booking, 3277 03:32:21,862 --> 03:32:24,976 OpenAI found a recorded device hidden in a conference room. That device, 3278 03:32:24,992 --> 03:32:28,812

the US court filing confirmed, was there for at least multiple days 3279 03:32:28,844 --> 03:32:32,664 and had been in place during a top-secret meeting with US government officials. 3280 03:32:33,563 --> 03:32:36,387 In the days surrounding the time period that the device was installed, 3281 03:32:36,692 --> 03:32:40,207 OpenAI executives met met with government officials in that very room. The filing said 3282 03:32:40,544 --> 03:32:44,059 defendant's conduct could pose a danger to the national security of the United States. 3283 03:32:44,814 --> 03:32:49,035 In their complaint, the U.S. alleged that Alfrazan used the OpenAI productivity 3284 03:32:49,131 --> 03:32:52,646 agent to research which conference rooms might be the best exploited for his 3285 03:32:52,678 --> 03:32:56,996 spying. Logs suggested that he used his wife's personal ChatGPT 3286 03:32:57,496 --> 03:33:02,067 account to research ways to modify his covert recording devices so they would auto-record 3287 03:33:02,567 --> 03:33:06,854 when detecting speech and allow remote access controls. After his arrest, Al-Fazan refused 3288 03:33:06,870 --> 03:33:10,324 to speak to the FBI, requested his attorney. However, his wife 3289 03:33:10,340 --> 03:33:13,828 confirmed that she had seen her ChatGPT account open on his laptop. 3290 03:33:14,280 --> 03:33:17,890 When asked about her plan to move back to Saudi Arabia, the government's filings said— 3291 03:33:18,390 --> 03:33:20,711 what, she threw him under the bus, Alex? I didn't know she threw him under 3292 03:33:21,211 --> 03:33:24,986 the bus. What, the fact that she ditched 3293 03:33:25,486 --> 03:33:28,501 him and, and moved to a different continent didn't— Well, listen, 3294 03:33:28,646 --> 03:33:32,206 listen, hey now, hey now, if you were, if you were his place, 3295 03:33:32,270 --> 03:33:35,458 you would tell her to take the son and leave and get out of here, 3296 03:33:35,586 --> 03:33:38,387 would you not? Would you not encourage your wife, or would you say, no, 3297 03:33:38,420 --> 03:33:41,608 stay here with me? Well, let's just say 3298 03:33:42,108 --> 03:33:45,418 that the totality of the evidence seems to be suggesting other things. 3299 03:33:45,644 --> 03:33:48,260 Yeah. I guess now he can throw her under the bus, 3300 03:33:49,018 --> 03:33:51,358 and that would be a substantially different game. 3301 03:33:54,037 --> 03:33:58,909 The government— I got to say, I'm shocked that somehow 3302 03:33:59,329 --> 03:34:02,982 he wasn't arrested in Las Vegas. I suppose. So wherever 3303 03:34:03,482 --> 03:34:06,701 he was arrested, they have discovered the technology of 3304 03:34:06,879 --> 03:34:10,403 keeping a suspect imprisoned and not letting 3305 03:34:10,500 --> 03:34:13,622 him— giving him his passport and just say, 3306 03:34:13,638 --> 03:34:17,274 we'll need you back, but you feel free to do international 3307 03:34:17,774 --> 03:34:18,369 travel wherever. 3308 03:34:21,598 --> 03:34:24,661 The government's motion for Al-Fazan's detention included more details 3309 03:34:24,677 --> 03:34:28,090 than previously reported from the search of Al-Fazan's house. In addition to Finding 3310 03:34:28,122 --> 03:34:32,008 recording devices similar to the ones planted in OpenAI offices, the FBI 3311 03:34:32,056 --> 03:34:35,107 allegedly found evidence that AlphaZone had planted at least one other bug. 3312 03:34:35,653 --> 03:34:39,282 Photos on AlphaZone's phone depicted at least one additional concealed recording device, 3313 03:34:39,667 --> 03:34:43,216 as well as what appeared to be input/output panels on televisions and other 3314 03:34:43,716 --> 03:34:46,234 electronic devices in various locations in OpenAI buildings, 3315 03:34:46,909 --> 03:34:50,136 including another conference room where sensitive meetings frequently occurred. 3316 03:34:50,184 --> 03:34:53,315 The filing said OpenAI and the U.S. government are both aware that. 3317 03:34:53,380 --> 03:34:57,406 Saudi Arabia had dedicated this entire year to advancing its AI through 3318 03:34:57,906 --> 03:35:01,337 an effort called Vision 2030. Sam Altman himself 3319 03:35:01,837 --> 03:35:04,898 joined Donald Trump at a summit where the Saudi government's AI push was celebrated. 3320 03:35:06,086 --> 03:35:09,487 However, the US is very guarded about American frontier labs. 3321 03:35:09,535 --> 03:35:13,449 The FBI therefore found it particularly concerning that Alfazan's phone showed WhatsApp 3322 03:35:13,530 --> 03:35:16,738 logs where a member of the Saudi royal family seemed to be 3323 03:35:17,238 --> 03:35:20,829 recruiting Alfazan. Those logs date back to 2025, the same year 3324 03:35:21,329 --> 03:35:24,937 he started at OpenAI, and continued up to October 5th. They included 3325 03:35:25,437 --> 03:35:29,127 an exchange where Al-Fazan responded 100 to a message urging, the Saudi 3326 03:35:29,627 --> 03:35:33,349 needs you more than the United States. His iPhone browser 3327 03:35:33,445 --> 03:35:36,816 activity was similarly revealing. The U.S. alleged on October 3328 03:35:37,316 --> 03:35:40,701 4th Al-Fazan accessed a folder called Saudi Government Docs. The FBI found the U.S. 3329 03:35:40,749 --> 03:35:44,409 did not make public what was in the folder, but they did explain what Al-Fazan 3330 03:35:44,521 --> 03:35:47,731 did after accessing this folder, which suggests he may 3331 03:35:48,231 --> 03:35:52,740 have been aiding the Saudi government. Logs show that Al-Fazan used OpenAI's 3332 03:35:53,240 --> 03:35:56,270 productivity agent to find out who from OpenAI could be suitable to attend high-level information 3333 03:35:56,319 --> 03:35:59,705 meetings with ministers in Saudi Arabia. List 10 people 3334 03:35:59,914 --> 03:36:02,065 and why each, ideally high levels. 3335 03:36:02,963 --> 03:36:06,639 Although Afrasan only faces a maximum sentence of 5 years for illegally 3336 03:36:06,655 --> 03:36:09,897 recording OpenAI's meetings, the government said

it expects it may raise more 3337 03:36:10,397 --> 03:36:13,893 serious charges as evidence mounts. Here, the nature and circumstance 3338 03:36:14,393 --> 03:36:17,907 of the offenses are indisputably serious, the filings said the defendant went to 3339 03:36:18,407 --> 03:36:22,451 significant lengths to create customized encrypted recording devices to secretly record 3340 03:36:22,467 --> 03:36:26,386 highly sensitive meetings involving senior OpenAI personnel and executives, 3341 03:36:26,739 --> 03:36:30,657 high-level business counterparts, and government officials, all while 3342 03:36:30,738 --> 03:36:35,218 communicating with individuals of significance in Saudi Arabia and pursuing opportunities 3343 03:36:35,395 --> 03:36:37,659 in and making plans to travel to Saudi Arabia. 3344 03:36:39,104 --> 03:36:42,862 The US said it's possible that Al-Fazl was acting on his own, but suggested that 3345 03:36:43,362 --> 03:36:46,090 it didn't matter if the government could prove he planted bugs while working for the 3346 03:36:46,590 --> 03:36:50,265 Saudi government. Quote, whether the defendant wanted to record OpenAI meetings to enrich himself 3347 03:36:50,765 --> 03:36:54,359 by enhancing his own business prospects or to provide aid to a foreign government or 3348 03:36:54,859 --> 03:36:58,261 both, his conduct is criminal and it presents a danger to innovation in Silicon 3349 03:36:58,293 --> 03:37:01,874 Valley and concern for the national security of the United States, the US argued, 3350 03:37:02,371 --> 03:37:06,048 asking the courts to deem Al-Farzan a flight risk. The US argued that Al-Farzan 3351 03:37:06,080 --> 03:37:09,484 has every incentive to flee and the use of information he allegedly stole 3352 03:37:09,517 --> 03:37:13,129 at OpenAI to possibly emerge as a talent at advancing Saudi Arabia's 3353 03:37:13,209 --> 03:37:16,805 AI goals. He faces prison time in the United States, the filing said. 3354 03:37:17,239 --> 03:37:20,482 His wife and son are now in Saudi Arabia and have no plans to return. 3355 03:37:20,851 --> 03:37:23,933 He has the resources to flee. He has the connections he needs not only to 3356 03:37:23,982 --> 03:37:27,514 flee, but to build a new life in Saudi Arabia. 3357 03:37:28,653 --> 03:37:31,463 So there you go, Alex. There's our update this week on Mr. 3358 03:37:31,591 --> 03:37:35,155 Al-Fazan and his espionage, corporate espionage, 3359 03:37:35,219 --> 03:37:38,703 or government espionage. I have to protest here 3360 03:37:39,410 --> 03:37:43,002 the completely unfair implication of the Saudis 3361 03:37:43,034 --> 03:37:47,496 in this attack on American infrastructure, 3362 03:37:48,453 --> 03:37:51,678 plus the previous one that you mentioned. I mean, at this rate, 3363 03:37:52,001 --> 03:37:55,219 we're going to accuse them of everything. What's next, 9/11? 3364 03:38:01,787 --> 03:38:05,011 How many are you racking up today of those 3365 03:38:05,449 --> 03:38:08,673 quacks there, Alex? Come

on. I think this may be 3366 03:38:08,690 --> 03:38:09,467 a personal record. 3367 03:38:13,200 --> 03:38:17,571 So yeah, I mean, listen, he's already— listen, 3368 03:38:18,069 --> 03:38:21,893 I don't— if the way that this article is reading, it's suggestive, 3369 03:38:22,022 --> 03:38:25,653 Alex, that he was approached after he acquired this job 3370 03:38:26,153 --> 03:38:30,409 and not necessarily before, and that he wasn't acquiring this job as part of a 3371 03:38:30,457 --> 03:38:33,719 plan by the Saudis, but that they found out he was employed 3372 03:38:34,635 --> 03:38:38,354 and tried to recruit him for specific, uh, tasks perhaps. 3373 03:38:38,581 --> 03:38:42,185 Um, uh, but, uh, um, I have a random 3374 03:38:42,685 --> 03:38:45,759 demographics question for you. All right. Um, 3375 03:38:46,276 --> 03:38:50,228 before this man was arrested, how many Saudi employees were working at OpenAI, 3376 03:38:50,309 --> 03:38:53,960 do you think? Zero. Well, before he was 3377 03:38:54,460 --> 03:38:58,208 arrested. One. One, exactly. 3378 03:38:58,321 --> 03:39:02,030 There was one, one Saudi 3379 03:39:08,083 --> 03:39:11,261 Right, so the pickings were slim and they didn't pick someone who was, uh, 3380 03:39:11,469 --> 03:39:15,401 apparently— sorry, 3381 03:39:15,481 --> 03:39:19,252 they, they didn't, they didn't pick the, the, the Tom Cruise, uh, you know, 3382 03:39:19,332 --> 03:39:22,253 Mission: Impossible spy guy, uh, that, uh, 3383 03:39:22,606 --> 03:39:25,110 that they could have. They, they had to pick— they had to go with the 3384 03:39:25,610 --> 03:39:27,838 team that they had, right? They had to fight the war with the— This is 3385 03:39:27,902 --> 03:39:30,898 more Mr. Bean than Than Mission: Impossible. 3386 03:39:30,947 --> 03:39:33,187 Yes, yes, very much Mr. Bean. 3387 03:39:36,994 --> 03:39:40,959 Now, uh, you know, I'm sure that they're— I'm sure that they were hoping 3388 03:39:40,991 --> 03:39:43,827 to get whatever they could. So whether it was technological information, 3389 03:39:44,830 --> 03:39:48,895 um, you know, insider secrets, uh, that could help them be competitive 3390 03:39:49,395 --> 03:39:52,153 on a global market or something like that, or even just to advance their own 3391 03:39:52,201 --> 03:39:55,487 stuff. Plus, as a bonus, hey, if there's any uh, you know, 3392 03:39:55,616 --> 03:39:58,952 government, government conversations that we can be privy to, 3393 03:39:59,016 --> 03:40:02,271 that'd be awesome as well. You know, I, I don't know if there was a 3394 03:40:02,320 --> 03:40:04,683 specific, um, 3395 03:40:05,265 --> 03:40:08,955 idea of what they could get through these 3396 03:40:08,988 --> 03:40:12,377 conversations, what they, what they were— maybe it was just hoping, you know, 3397 03:40:12,538 --> 03:40:16,732 why not? It is, it is fascinating to wonder what exactly. 3398 03:40:16,796 --> 03:40:19,624 I mean, because we have somebody on the inside who will tell you whatever he 3399 03:40:20,124 --> 03:40:23,601 knows, right? So, you know, whatever is told 3400 03:40:24,101 --> 03:40:27,640 to anyone at OpenAI, plus whatever he can snoop into without, 3401 03:40:27,720 --> 03:40:31,677 you know, raising too much suspicion, he could tell you. Right. Instead of that, 3402 03:40:31,758 --> 03:40:35,988 they went with like, let's bug the, you know, the most 3403 03:40:36,117 --> 03:40:40,062 secretive conference 3404 03:40:40,562 --> 03:40:44,120 rooms for, you know, a fishing expedition 3405 03:40:44,620 --> 03:40:48,606 into who knows what. You know, so for 3406 03:40:49,106 --> 03:40:52,662 example, uh, OpenAI fired 3 safety researchers last week, 3407 03:40:52,678 --> 03:40:56,429 I think. I don't know if you remember that story. Yeah. And it emerged finally 3408 03:40:56,622 --> 03:40:59,828 that— I mean, look, there's some speculation here, but, 3409 03:41:00,453 --> 03:41:03,723 uh, it sounds pretty plausible to me— that they believed, 3410 03:41:04,381 --> 03:41:07,747 um, that these 3 people leaked the fact that the latest model, 3411 03:41:07,907 --> 03:41:12,107 Astra, is a looped transformer. And they took that seriously enough to fire everybody. 3412 03:41:12,877 --> 03:41:16,182 who they found had a connection through 3413 03:41:16,682 --> 03:41:20,706 Signal. Like literally they just found an empty Signal room because they were using disappearing 3414 03:41:20,754 --> 03:41:24,396 messages to a specific person at MITRE 3415 03:41:24,925 --> 03:41:28,454 called Ryan Greenblatt, who they believed informed 3416 03:41:28,954 --> 03:41:32,401 the reporter who publishes information, right? Mm-hmm. Like that information is 3417 03:41:33,171 --> 03:41:36,139 important, but like not, okay, like I mean, 3418 03:41:36,155 --> 03:41:38,705 it's not, loop transformers are known, 3419 03:41:39,827 --> 03:41:43,450 Um, you know, it is a technology that a few companies have looked into. 3420 03:41:43,691 --> 03:41:47,153 The fact that OpenAI has gone into it, you know, 3421 03:41:47,378 --> 03:41:50,840 this is information— what I'm saying is like, if it's that important, somebody like 3422 03:41:50,904 --> 03:41:54,799 this guy would have known that, plus a bunch of other bits and pieces 3423 03:41:55,152 --> 03:41:58,999 that he could totally have been leaking without anyone, you know, without doing 3424 03:41:59,499 --> 03:42:02,926 anything as stupid as what he did. Which makes you wonder, uh, 3425 03:42:03,007 --> 03:42:07,146 what were they going for here? Like, what was the What is 3426 03:42:07,339 --> 03:42:10,527 the pot of gold at the end of the rainbow that goes 3427 03:42:10,544 --> 03:42:15,551 through putting a camera on 3428 03:42:16,051 --> 03:42:19,502 a smart TV and obscuring it with toilet paper? Right, right, right. 3429 03:42:20,050 --> 03:42:23,973 Yeah. I mean, there's a lot of questions. 3430 03:42:24,102 --> 03:42:27,554 And look, from the Saudi government's standpoint, 3431 03:42:27,651 -->

03:42:31,055 okay, guy gets caught, sort of his problem. 3432 03:42:32,042 --> 03:42:34,047 Not ours. What's gonna happen? 3433 03:42:35,559 --> 03:42:37,830 Donald Trump's gonna be a little bit upset with us. 3434 03:42:40,730 --> 03:42:44,138 Ooh. All of this is considered, you know, that level is considered, 3435 03:42:44,203 --> 03:42:47,239 you know, friendly jabbing amongst friends. 3436 03:42:47,530 --> 03:42:50,856 I think it's not exactly that big of a deal. Exactly. So this guy, 3437 03:42:51,373 --> 03:42:55,344 you know, he's screwed and the Saudi government's just 3438 03:42:55,844 --> 03:42:59,580 like, whoop, that plan didn't work. Back to the drawing board. 3439 03:42:59,650 --> 03:43:02,510 Now, if you were smart, you might arrest a couple of Americans and do a 3440 03:43:03,010 --> 03:43:06,770 hostage deal or give Trump another plane or whatever. 3441 03:43:08,390 --> 03:43:12,110 Usually that's how this is sorted out. Have you ever heard of 3442 03:43:12,610 --> 03:43:16,450 the movie Double Indemnity? Oh, 3443 03:43:16,950 --> 03:43:20,230 you've mentioned this before. No, no, I haven't watched it. All right. All right. 3444 03:43:20,730 --> 03:43:24,370 Well, the Walter Matthau movie. Yes. The insurance 3445 03:43:24,870 --> 03:43:28,423 agent, the life insurance agent. Um, he, 3446 03:43:28,472 --> 03:43:32,486 he starts to talk about how if you've worked in an environment long enough and 3447 03:43:32,502 --> 03:43:36,323 know all the rules, you start to— your mind starts to wander and think about 3448 03:43:36,355 --> 03:43:39,807 ways that with your knowledge you could exploit it, you know, 3449 03:43:40,032 --> 03:43:43,308 find, find ways of getting around certain things. And, and, 3450 03:43:43,484 --> 03:43:46,632 uh, this is one of those situations, Alex, 3451 03:43:46,696 --> 03:43:49,955 where with just a little bit of knowledge of your own industry, you would have 3452 03:43:49,987 --> 03:43:53,536 been able to perform a little bit better. Right. 3453 03:43:54,098 --> 03:43:57,277 He could have just simply subverted one of those 3454 03:43:57,389 --> 03:44:00,745 little tests that they do and had 3455 03:44:01,245 --> 03:44:04,117 an agent write on one of those fancy message boards that they're talking about all 3456 03:44:04,197 --> 03:44:08,163 their AI agents writing to and leave a message that says, hey, why don't 3457 03:44:08,663 --> 03:44:14,169 you hack one of these several people and spread this information to the Saudi government? 3458 03:44:14,265 --> 03:44:17,589 Do something where at least they would've gone, oh, they went rogue and contacted 3459 03:44:18,089 --> 03:44:21,325 the Saudis. Saudi government and they hacked, you know. 3460 03:44:21,551 --> 03:44:25,031 He was fired for being less competent than their AI agents. 3461 03:44:25,750 --> 03:44:29,173 That's the bar at OpenAI. It's like, 3462 03:44:29,542 --> 03:44:33,223 you gotta be as good as this, which is, this is being our latest model.

3463 03:44:33,833 --> 03:44:36,650 If you can't do the job that well, 3464 03:44:37,343 --> 03:44:40,822 well, we're sorry. The men in black are just gonna come and 3465 03:44:41,322 --> 03:44:44,604 take you away. An agent wrote on a message board that we had to hack 3466 03:44:45,104 --> 03:44:48,984 Sam Altman and to give his email, entire email 3467 03:44:49,162 --> 03:44:52,414 dump to the Saudi Arabian government for some reason. We were trying to get 3468 03:44:52,914 --> 03:44:56,549 to the bottom of where that came from. Oh, it was 3469 03:44:56,759 --> 03:45:00,578 actually, it was his wife's normal customer account somehow. 3470 03:45:00,643 --> 03:45:04,154 That, that they're so rogue, Alex. 3471 03:45:04,219 --> 03:45:07,286 They're so rogue. Yep. 3472 03:45:07,866 --> 03:45:11,249 Yep. Oh man. All right, 3473 03:45:11,878 --> 03:45:14,633 Alex, this is the last story I have on the stack. And of course, 3474 03:45:14,713 --> 03:45:18,629 after that, if anyone wants to comment, go ahead and request to speak, 3475 03:45:19,047 --> 03:45:22,727 that'll be your time to shine. And otherwise, Alex, 3476 03:45:22,743 --> 03:45:25,362 if you have anything that we have missed that you would like to discuss, 3477 03:45:25,378 --> 03:45:29,186 you can bring that up. Intel gets waiver after worries fuzzy 3478 03:45:29,315 --> 03:45:32,191 US ban on foreign-made robots may hit fabs. 3479 03:45:33,493 --> 03:45:37,253 US chip champion Intel was so confused by the US government's ban on 3480 03:45:37,269 --> 03:45:40,483 foreign-made robots that it requested and received a waiver, quote, 3481 03:45:40,563 --> 03:45:43,712 out of an abundance of caution. The tech company wanted to make sure that it 3482 03:45:44,212 --> 03:45:47,164 could continue using certain robotic vehicles in its U.S. chip fabs. 3483 03:45:47,951 --> 03:45:51,949 The uncertainty surrounding the Federal Communications Commission's ban on certain foreign-made 3484 03:45:51,965 --> 03:45:55,434 robots was revealed in the agency's decision to grant Intel's waiver 3485 03:45:55,934 --> 03:46:00,138 request. On October 22nd, the FCC published its order allowing 3486 03:46:00,234 --> 03:46:03,654 Intel to continue using the foreign-made robots in 3487 03:46:03,703 --> 03:46:06,930 U.S. semiconductor manufacturing facilities for a period of 1 year. 3488 03:46:07,235 --> 03:46:10,816 The robots in question are automated guided vehicles operating on tracks or 3489 03:46:10,848 --> 03:46:14,348 predefined paths while often requiring operator oversight, 3490 03:46:14,364 --> 03:46:18,426 according to Intel's website. They help to move materials and items around within 3491 03:46:18,474 --> 03:46:21,974 warehouses and factory floors. Such robotic vehicles are typically 3492 03:46:21,990 --> 03:46:25,955 manufactured by Japanese companies like Murata Machinery and Daifuku 3493 03:46:26,164 --> 03:46:30,049 Co., according to Bloomberg News. But the exact 3494 03:46:30,549 --> 03:46:34,127 types of robotic vehicles are used by Intel were only ident— uh, 3495 03:46:34,448 --> 03:46:37,595 were only identified in a confidential appendix that accompanied its 3496 03:46:37,643 --> 03:46:41,764 request. Now, Alex, I just like to point out that despite my horrendous pronunciation 3497 03:46:42,454 --> 03:46:46,894 of French names last week, Japanese stuff 3498 03:46:47,135 --> 03:46:50,437 is so much easier to do because everything is much more, 3499 03:46:51,687 --> 03:46:55,551 much more readable. Like it doesn't matter what 3500 03:46:56,051 --> 03:46:59,254 it is because they use a, they use a syllabary. And so everything 3501 03:46:59,302 --> 03:47:02,957 is effectively phonetic. If you understand how the pronunciation 3502 03:47:02,989 --> 03:47:05,742 rules go, it just, it's easy. Easy. That's right. 3503 03:47:05,807 --> 03:47:09,439 That's right. They don't they don't use the the Latin 3504 03:47:09,939 --> 03:47:12,926 alphabet to begin with, so when they transliterate, it becomes phonetic. 3505 03:47:13,876 --> 03:47:16,227 There's no weird friendship going on. Yes. 3506 03:47:16,790 --> 03:47:20,027 Daifu Coco. So much easier. 3507 03:47:20,600 --> 03:47:24,665 The FCC's ban mainly covers new models of mobile robots like humanoid robots 3508 03:47:25,165 --> 03:47:29,554 and robot vacuums, and still allows for the most for the import of robots previously 3509 03:47:29,651 --> 03:47:33,298 approved for sale in the United States. But the agency ruled. Seem to restrict 3510 03:47:33,798 --> 03:47:37,505 major modifications to previously authorized devices. Intel's waiver request 3511 03:47:38,005 --> 03:47:41,584 stated that it is unclear whether its robotic vehicles even qualify as robots 3512 03:47:41,889 --> 03:47:45,790 covered by the FCC ban, but it said it sought relief just in case. 3513 03:47:45,903 --> 03:47:48,632 The tech company also received approval to make certain hardware, 3514 03:47:48,649 --> 03:47:52,181 software, and firmware changes that would help ensure the continued operation 3515 03:47:52,681 --> 03:47:56,613 and maintenance of the robotic vehicles. The tech company described the robotic vehicles 3516 03:47:56,902 --> 03:48:00,384 as being critical to its domain domestic semiconductor manufacturing operations 3517 03:48:00,884 --> 03:48:04,426 and argued that the ability to maintain the robots was crucial to avoiding any 3518 03:48:04,458 --> 03:48:08,517 disruptions to chip production. It also said there are no U.S.-made alternatives 3519 03:48:08,549 --> 03:48:12,111 that meet the company's requirements. The FCC approval came 3520 03:48:12,159 --> 03:48:16,089 within a week of when Intel filed its waiver request on October 16th, 3521 03:48:16,490 --> 03:48:20,549 and that's not too surprising given the importance of Intel to the Trump administration's plans 3522 03:48:20,581 --> 03:48:24,126 for reshoring chip manufacturing. Not to mention that the U.S. 3523 03:48:24,190 --> 03:48:27,864 government

took an $8.9 billion equity stake 3524 03:48:27,944 --> 03:48:31,697 in the chip company. Meanwhile, the U.S. robotics industry at large 3525 03:48:31,730 --> 03:48:35,162 is still grappling with the consequences of the FCC ban on foreign-made 3526 03:48:35,662 --> 03:48:39,028 robots, given that China is the world's leading supplier of robots and robotic 3527 03:48:39,528 --> 03:48:43,310 components. Ours previously talked with one U.S. robotics distributor 3528 03:48:43,599 --> 03:48:46,855 that has pivoted from importing China's humanoid robots to opening a U.S. 3529 03:48:46,951 --> 03:48:49,614 manufacturing facility for robots. 3530 03:48:50,031 --> 03:48:54,025 Some analysts have warned the U.S. government's attempt to boost domestic production via 3531 03:48:54,169 --> 03:48:57,873 restrictions on foreign-made robots could backfire for U.S. industry. A report 3532 03:48:58,373 --> 03:49:01,928 by the market intelligence firm International Data Corporation looks forward to 2030 3533 03:49:02,440 --> 03:49:06,046 and estimates that the FCC rule could reduce US revenue from deploying humanoid 3534 03:49:06,546 --> 03:49:10,004 robots by 58% compared to pre-restriction expectations. And Alex, 3535 03:49:10,453 --> 03:49:14,571 as you've pointed out in the past, certain things are so difficult 3536 03:49:14,603 --> 03:49:18,353 to figure out that you have talked about one person with a project of trying 3537 03:49:18,385 --> 03:49:22,861 to create a scrubber for grill 3538 03:49:23,542 --> 03:49:26,993 inside, you know, purely using stuff sourced from the United 3539 03:49:27,025 --> 03:49:30,443 States, and that they have been unsuccessful in producing a wire scrubber 3540 03:49:30,943 --> 03:49:34,993 brush for it. That is correct. It is a chainmail scrubber 3541 03:49:35,493 --> 03:49:38,809 brush, to be clear. But nevertheless, 3542 03:49:39,303 --> 03:49:40,718 yes, it's incredible. 3543 03:49:42,560 --> 03:49:45,760 Now, Alex, this is the first time I think in many, many, many months that 3544 03:49:46,260 --> 03:49:50,043 we've actually successfully gotten through my stack. Now, that's partially because because I was 3545 03:49:50,091 --> 03:49:53,286 somewhat delayed in starting due to my medical event. 3546 03:49:53,832 --> 03:49:56,899 Uh, but, but I still— Can I inject, 3547 03:49:57,043 --> 03:49:59,901 uh, can I inject, uh, a little story here? 3548 03:50:00,656 --> 03:50:04,236 Yes. Okay. Over 1,000 mathematicians have now 3549 03:50:04,285 --> 03:50:08,106 joined the Association for Human Mathematics, committing to not publish 3550 03:50:08,363 --> 03:50:11,799 AI-generated math papers, not work with Frontier Labs, 3551 03:50:12,377 --> 03:50:16,407 and 654 of them went even further to not 3552 03:50:16,907 --> 03:50:19,230 even use AI in the research All right. 3553 03:50:21,415 --> 03:50:24,490 We will see how they do. I mean, that's fine, right? 3554 03:50:25,860 --> 03:50:29,238 That's— this is, this is fine. And perhaps, 3555 03:50:29,529 --> 03:50:32,050 perhaps they will be better off for it. There's one reason I don't understand. 3556 03:50:32,842 --> 03:50:35,509 The first tweet is a simple image of Darwin. 3557 03:50:37,453 --> 03:50:40,233 I'm not sure what is being conveyed here, 3558 03:50:41,251 --> 03:50:45,039 but something about an award, I assume. Well, you do understand 3559 03:50:45,539 --> 03:50:48,937 that Darwin has a beard in most images, right? Correct. 3560 03:50:49,241 --> 03:50:52,930 Okay, do you understand how authoritative beards 3561 03:50:53,058 --> 03:50:56,201 are whenever they're— Very. 3562 03:50:56,683 --> 03:51:00,820 They're the most authoritative. You understand how Ayn Rand 3563 03:51:01,398 --> 03:51:03,755 is one of the most successful, uh, 3564 03:51:04,381 --> 03:51:07,748 philosophers of modern times when it comes to their impact 3565 03:51:07,813 --> 03:51:11,508 on the world, but they are laughed at as philosophers because 3566 03:51:11,540 --> 03:51:14,940 they don't have a compelling beard. Ayn Rand does not have a 3567 03:51:15,440 --> 03:51:17,981 compelling beard. I don't want to judge, 3568 03:51:18,061 --> 03:51:21,371 but— She was not as hairy as Marx, 3569 03:51:21,531 --> 03:51:23,861 sir. No, no, 3570 03:51:23,942 --> 03:51:25,870 that is a high bar. 3571 03:51:27,536 --> 03:51:29,213 So, you know, anyway, yeah, so, um, 3572 03:51:30,858 --> 03:51:33,890 it's, it's interesting. This is, this is the whole, you know, 3573 03:51:33,922 --> 03:51:38,082 we didn't touch on the whole AI math crash out. 3574 03:51:38,569 --> 03:51:41,667 But all this is very— the AHM, by the way, is the 3575 03:51:42,167 --> 03:51:45,990 organization started by Terence Tao. So, okay, this is 3576 03:51:46,022 --> 03:51:49,798 not some like random— There was nothing on my radar this week 3577 03:51:50,298 --> 03:51:54,834 that was major from the math community. I kind of did see that 3578 03:51:55,254 --> 03:51:59,514 announcement you made, but it didn't strike me as being majorly 3579 03:51:59,611 --> 03:52:04,420 different than the last couple of weeks. Is there some aspect of it that 3580 03:52:04,920 --> 03:52:07,594 you find— No, no, I'm just keeping everybody warm to the fact that yes, 3581 03:52:07,658 --> 03:52:11,479 the mathematicians are still crashing out. I don't know 3582 03:52:11,495 --> 03:52:16,664 what state of the denial stage 3583 03:52:17,164 --> 03:52:20,340 of grief they're in, but it's getting worse. 3584 03:52:20,726 --> 03:52:24,964 There are people on the radar out there who have 3585 03:52:24,980 --> 03:52:29,298 been trying to figure out what path 3586 03:52:29,314 --> 03:52:32,816 they're going to And one person I saw 3587 03:52:33,265 --> 03:52:36,527 in the mathematical community has decided 3588 03:52:36,559 --> 03:52:40,784 that he's going to start taking his students down 3589 03:52:40,880 --> 03:52:44,575 the course of computer science and mathematics 3590 03:52:45,075 --> 03:52:49,523

intersections and begin to work on having all of his students 3591 03:52:50,535 --> 03:52:53,651 build LLMs and train them 3592 03:52:54,151 --> 03:52:58,236 on mathematical content so that they can better analyze and understand what the 3593 03:52:58,736 --> 03:53:03,727 LLMs are doing with the mathematics that they are ingesting and 3594 03:53:03,744 --> 03:53:07,705 being— So doing the interpretability. Yeah. And that 3595 03:53:08,661 --> 03:53:12,176 he is also going to begin himself studying and then 3596 03:53:13,747 --> 03:53:17,749 teaching with his students algorithmic 3597 03:53:17,798 --> 03:53:21,412 verifications of mathematical formulas and building 3598 03:53:21,509 --> 03:53:24,638 your own computer algebra systems and stuff like that so that, 3599 03:53:25,391 --> 03:53:29,545 you know, without knowing what direction it's going to go, these things 3600 03:53:29,594 --> 03:53:32,898 are going to maintain their importance. And so you might as well know 3601 03:53:33,732 --> 03:53:37,918 a whole lot about that, which I absolutely agree 3602 03:53:38,418 --> 03:53:41,446 with. So it sounds a lot more reasonable than some of the other takes I've 3603 03:53:41,946 --> 03:53:44,895 seen. I'll tell you that much. Yeah. I mean, you know, 3604 03:53:45,119 --> 03:53:48,166 I've— there's the one thing I did not put on the stack this 3605 03:53:48,666 --> 03:53:52,241 week is that, you know, there's been plenty of channels over the years 3606 03:53:53,220 --> 03:53:57,090 with tech advice, whether it's programming advice or other 3607 03:53:57,590 --> 03:54:00,975 types of stuff where it's like, they'll do small, 3608 03:54:01,570 --> 03:54:05,166 maybe 30 to 45-minute videos of a how-to 3609 03:54:05,343 --> 03:54:08,972 on something, how to set up your own X server, something server, 3610 03:54:09,550 --> 03:54:12,665 how to dispatch some other product or blah, 3611 03:54:12,681 --> 03:54:16,309 blah, blah, blah, blah, how to configure XYZ, how to program 3612 03:54:16,358 --> 03:54:19,119 some shell script, blah, blah, blah. 3613 03:54:19,746 --> 03:54:23,834 And a couple of them have been sort of pausing 3614 03:54:24,188 --> 03:54:28,485 that sort of endeavor and going through comments 3615 03:54:28,598 --> 03:54:31,335 that they're getting in their own feeds and stuff like that, 3616 03:54:31,431 --> 03:54:35,231 talking about difficulties that even 3617 03:54:35,731 --> 03:54:38,068 seasoned developers are having in the environment finding work. 3618 03:54:40,434 --> 03:54:43,946 It does not appear from my knowledge, looking at 3619 03:54:43,962 --> 03:54:48,127 broad statistics, that there has been an incredible 3620 03:54:48,239 --> 03:54:52,208 drop-off in employment in tech, 3621 03:54:52,835 --> 03:54:56,016 let's call them programming areas and software development. 3622 03:54:56,273 --> 03:55:00,402 But there is— it's a non-zero drop-off. So like 5% 3623 03:55:00,451 --> 03:55:04,275 reduction. I'm seeing, I'm seeing some numbers of that. 3624 03:55:04,468 --> 03:55:08,035 Yeah, it's a soft market. It's not a crashing market, 3625 03:55:08,099 --> 03:55:11,280 but it's getting— Yeah. Um, but there's, there's apparently 3626 03:55:11,697 --> 03:55:15,849 an increasing desperation in those who have been on the early 3627 03:55:16,618 --> 03:55:18,911 chopping blocks. Yeah. 3628 03:55:19,888 --> 03:55:24,121 That what to do is there. Now, the mathematics 3629 03:55:24,621 --> 03:55:28,770 community, as I've defended them in the last couple weeks, sort of given them 3630 03:55:29,186 --> 03:55:32,793 a lot of extra sympathy, they don't 3631 03:55:33,293 --> 03:55:38,917 necessarily face exactly the same hiring pressures that the software people 3632 03:55:39,417 --> 03:55:41,977 Because lots of them are tenured and whatnot. Yeah, lots of them. I mean, 3633 03:55:41,993 --> 03:55:45,599 now you can feel a lot more additional sympathy for the people who are just 3634 03:55:45,615 --> 03:55:48,740 getting their PhDs or something going like, well, what the hell do I do now? 3635 03:55:49,605 --> 03:55:53,035 Like, oh no. Um, but I, I genuinely 3636 03:55:53,067 --> 03:55:56,384 think that there, there's going to be a lot of demand for 3637 03:55:56,736 --> 03:56:00,070 that kind of, uh, brainpower at the AI companies, uh, 3638 03:56:00,438 --> 03:56:03,661 counterintuitively enough. But they 3639 03:56:04,161 --> 03:56:07,354 might not get to do, you know, the thing that they wanted to do. 3640 03:56:07,499 --> 03:56:10,309 Yeah. At least not on the job. You're referring to the math people or the 3641 03:56:10,809 --> 03:56:14,548 software people? The math people. Yeah. The software 3642 03:56:14,564 --> 03:56:17,534 people, I don't know, man. Learn the math, 3643 03:56:17,679 --> 03:56:21,147 I guess. Sorry. Well, when I talked to our friend 3644 03:56:21,647 --> 03:56:24,679 Steve Patterson a couple years ago, and I refer to this one just because it's 3645 03:56:25,179 --> 03:56:29,232 my earliest on-the-record sort of podcast talking about all of these things, things several years 3646 03:56:29,426 --> 03:56:32,954 before they actually became really true, just because I was on 3647 03:56:32,986 --> 03:56:36,448 the record. I pointed out that while 3648 03:56:36,561 --> 03:56:40,987 there is the old laissez-faire observation 3649 03:56:41,263 --> 03:56:44,805 that the incandescent 3650 03:56:44,837 --> 03:56:48,456 bulb replaced all the candle makers and the car replaced 3651 03:56:48,956 --> 03:56:52,923 all the horse and buggy people and that things are fine, that it's not necessarily 3652 03:56:52,956 --> 03:56:56,650 fine for real human beings in the short 3653 03:56:57,150 --> 03:57:00,977 term as the transitionary period occurs, and that there's a lot of 3654 03:57:01,477 --> 03:57:04,839 room for trouble for those people to have a lot 3655 03:57:05,339 --> 03:57:08,973 of disruptive effects in their lives and a lot of soul searching and 3656 03:57:09,005 -->

03:57:13,428 periods of despair that can really have significant 3657 03:57:13,989 --> 03:57:17,290 strong negative impacts on real individuals' lives. And so that 3658 03:57:17,963 --> 03:57:21,392 the blasé, it'll all be fine sort 3659 03:57:21,892 --> 03:57:25,001 attitude is maybe correct on a long enough horizon, but it ignores a whole 3660 03:57:25,049 --> 03:57:29,081 lot of pain that people might go through. And I think that there 3661 03:57:29,581 --> 03:57:33,422 should be a real— there is a real need 3662 03:57:34,005 --> 03:57:37,996 for entrepreneurial people to think 3663 03:57:38,077 --> 03:57:42,852 about the current glut in layoffs 3664 03:57:43,610 --> 03:57:46,800 and talent that might be accessible that may 3665 03:57:46,880 --> 03:57:51,593 not be able to be repurposed immediately when there's just these low-cost LLM 3666 03:57:52,093 --> 03:57:55,360 agents available to do certain jobs. That might just be the new environment, but that 3667 03:57:55,860 --> 03:58:00,313 doesn't mean that the talent has no use. And hopefully 3668 03:58:00,813 --> 03:58:04,385 there are people looking for opportunities in that scenario to, 3669 03:58:05,859 --> 03:58:09,306 let's call it, lead the way to whatever is supposed to be next, 3670 03:58:09,787 --> 03:58:14,243 because not everyone is going to have the position to come 3671 03:58:14,275 --> 03:58:16,790 up with the idea or see it to fruition. 3672 03:58:17,255 --> 03:58:20,733 So they're going to be more desperate to have 3673 03:58:20,765 --> 03:58:24,130 someone create that for them and show them what the next thing forward 3674 03:58:24,630 --> 03:58:28,200 is. Because there's going to be too many people disrupted in the 3675 03:58:28,700 --> 03:58:31,982 near future that it's going to be too much confusion. And all that's going 3676 03:58:32,482 --> 03:58:35,780 to put pressure on is the political class to do something, and their solutions are 3677 03:58:36,280 --> 03:58:39,918 going to be garbage. We don't want them to be the ones solving it, 3678 03:58:41,266 --> 03:58:45,054 you know? Right. I don't know. Do you have any immediate 3679 03:58:45,554 --> 03:58:47,574 thoughts on, on what the, those, those, uh, 3680 03:58:48,328 --> 03:58:51,554 recently unemployed, laid-off software developers should 3681 03:58:52,054 --> 03:58:55,823 be retargeting their efforts towards? Because it can't all be the same thing. That's going 3682 03:58:56,323 --> 03:59:00,028 to crowd out whatever their ideas are. As a not so recently 3683 03:59:00,061 --> 03:59:03,848 unemployed software person that is not necessarily the, 3684 03:59:05,020 --> 03:59:08,727 um, you know, uh, burning with 3685 03:59:09,227 --> 03:59:12,547 a desire to find my next, my next thing. Um, 3686 03:59:13,285 --> 03:59:16,511 I, you know, I'm probably not well suited to answer, but I, 3687 03:59:17,715 --> 03:59:20,844 I, I, I, I do think this is a real problem. 3688 03:59:21,149 --> 03:59:24,455 Um, and that we will have

to figure this out, especially if it becomes a, 3689 03:59:24,696 --> 03:59:27,312 you know, a widespread issue. Um, 3690 03:59:28,387 --> 03:59:32,481 it is the sort of thing that can can make 3691 03:59:32,885 --> 03:59:34,403 a lot of things go sideways. 3692 03:59:37,077 --> 03:59:40,630 Whatever AI adoption 3693 03:59:41,130 --> 03:59:45,211 does, it really can't be allowed to lead 3694 03:59:45,711 --> 03:59:50,941 to widespread unemployment or things will go south pretty 3695 03:59:51,441 --> 03:59:55,593 quickly, I think. Yeah, that desperation will increase faster 3696 03:59:56,093 --> 03:59:58,682 than you can imagine. Um, so that's, 3697 03:59:59,887 --> 04:00:03,308 that's pretty important, I think. Um, uh, Alex, do you have any other 3698 04:00:03,324 --> 04:00:06,954 stories that you'd like to tackle before we put the tail end 3699 04:00:07,454 --> 04:00:11,178 on this episode here? No, no, I just, uh, I just— that one 3700 04:00:11,210 --> 04:00:14,423 was somehow on my feed and I wanted to, uh, 3701 04:00:14,905 --> 04:00:17,748 just update everyone that on the, on the state of mathematics. 3702 04:00:19,000 --> 04:00:21,940 And that reminds me of the, uh, the early early, 3703 04:00:22,602 --> 04:00:26,945 early, uh, culture war, uh, things back in the 2015 3704 04:00:26,993 --> 04:00:30,759 eras, uh, where it was, uh, Men Going Their Own 3705 04:00:30,888 --> 04:00:33,755 Way, where I guess they had decided that they were, they were, uh, 3706 04:00:33,803 --> 04:00:37,199 shrugging off all women because they were too, uh, 3707 04:00:37,907 --> 04:00:41,032 too complicated. They don't even have to find a new acronym. 3708 04:00:41,467 --> 04:00:43,655 They can just be the MGTOWs. 3709 04:00:45,377 --> 04:00:48,930 Mathematicians Going Their Own Way. Yeah, yeah, 3710 04:00:49,829 --> 04:00:53,808 we're not gonna— we're not gonna play that game. No, no AIs 3711 04:00:53,985 --> 04:00:56,777 for us. Well, it's like, okay, well, 3712 04:00:57,611 --> 04:01:00,949 listen, listen, I'm even more diehard than 3713 04:01:01,045 --> 04:01:04,223 you. I'm not gonna use a calculator. I'm only gonna do things 3714 04:01:04,271 --> 04:01:08,042 by hand. That's right. I'm purer than thou. And you know 3715 04:01:08,122 --> 04:01:11,861 what? I'm not even gonna use numbers. I'm just gonna compute with vibes. 3716 04:01:12,134 --> 04:01:16,258 All right. And I'm going to manually take the limit everywhere 3717 04:01:16,290 --> 04:01:19,709 I go. I'm going to write out all of those sequences by hand until I 3718 04:01:19,837 --> 04:01:23,064 get to infinity, goddammit. I'm gonna return to 3719 04:01:23,273 --> 04:01:24,349 Roman numeral. 3720 04:01:24,830 --> 04:01:28,315 Alright. No zeros, 3721 04:01:28,363 --> 04:01:31,702 no newfangled zeros for me. That's a, that's a good 3722 04:01:31,718 --> 04:01:35,218 place to tail-end that thing, Alex. I think I did okay, all things 3723 04:01:35,251 --> 04:01:37,771 considered. I'm hopefully— Oh my God, Gator, 3724 04:01:38,205 --> 04:01:41,576 please take care of yourself. But perhaps not by 3725 04:01:42,076 --> 04:01:45,652 going to the eventual room again. The eventual 3726 04:01:45,767 --> 04:01:46,781 room. I don't know. 3727 04:01:49,458 --> 04:01:52,823 Um, yeah, I don't— I hope your, your painkillers don't die 3728 04:01:53,323 --> 04:01:56,044 out. I'm, I mean, I'm, you know, I'm hoping you're, you're getting better, but you're 3729 04:01:56,544 --> 04:01:59,934 not. You haven't allowed yourself to. I think it's all right. 3730 04:02:00,226 --> 04:02:04,346 I think it's all right right now. It's, it's, um, moderate pain and, 3731 04:02:04,362 --> 04:02:07,493 uh, mixed with some queasiness. It's not as— it's definitely not what it 3732 04:02:07,510 --> 04:02:11,720 was, um, Thursday, at least. Like Thursday was horrific. 3733 04:02:12,220 --> 04:02:15,120 That was like the— that is like literally the part where you're like, am I— 3734 04:02:15,620 --> 04:02:19,100 am I dying? Is something killing me? Like, what's happening here? 3735 04:02:19,200 --> 04:02:22,580 Did my entire insides explode and I'm like 3736 04:02:23,080 --> 04:02:26,620 hemorrhaging blood inside? What's going on? Like, that was the type of like, what's happening 3737 04:02:27,120 --> 04:02:29,620 that I was worried about, you know? Um, 3738 04:02:30,120 --> 04:02:33,040 but, uh, yeah. So hopefully you'll make it through this dumb week. 3739 04:02:33,420 --> 04:02:37,230 Yes, we shall see. We shall see. 3740 04:02:37,500 --> 04:02:40,460 All right, Gator, take care, man. All right, everybody have a wonderfully dumb week and