WEBVTT
00:00:01.919 --> 00:00:06.480
This June was supposed to be the start of Hot AI summer.
00:00:06.719 --> 00:00:19.519
That's because earlier this month, Anthropic released its Mythos 5 and Fable 5 models to the general public, and we were supposed to have an avalanche of other Frontier models released to all.
00:00:19.679 --> 00:00:28.000
Google was slated to release Gemini 3.5 Pro, and OpenAI was on board to drop GPT 5.6 all this month.
00:00:28.160 --> 00:00:40.079
And although we technically did kind of get the GPT-5 release from OpenAI late Friday, it's not exactly the hot AI summer we were promised.
00:00:40.240 --> 00:00:46.880
That's because Anthropic's heavy capabilities hypings of Mythos caused the US government to step in.
00:00:47.119 --> 00:00:55.920
And now we're at the end of June with no general availability to the promised frontier models from any of the big three companies.
00:00:56.079 --> 00:01:08.799
Although we do have some updated news because Mythos 5 and the new GPT 5.6 have started to roll out, but only if you work at a select few companies.
00:01:08.959 --> 00:01:12.079
So that's the big AI news of the week.
00:01:12.159 --> 00:01:15.040
But there's a lot more that we're going to get you caught up on.
00:01:15.200 --> 00:01:23.200
So if you don't have hours every single day trying to understand what's happening in the AI world and how it impacts your business, don't worry.
00:01:23.359 --> 00:01:25.120
That's what I do every single day.
00:01:25.200 --> 00:01:28.159
And then on Monday, we bring you the AI news that matters.
00:01:28.400 --> 00:01:29.760
So let's get into it.
00:01:29.840 --> 00:01:30.879
If you're new here, welcome.
00:01:30.959 --> 00:01:33.599
My name is Jordan Wilson, and this is Everyday AI.
00:01:33.760 --> 00:01:46.480
It's a daily live stream podcast and free daily news that are helping everyday business leaders like you and me not just keep up, but how we can use all of this information hidden between the headlines to get ahead and grow our companies and careers.
00:01:46.560 --> 00:01:50.319
So it starts here with the unedited, unscripted live stream podcast.
00:01:50.400 --> 00:01:56.079
But please make sure if you haven't already, subscribe to the podcast on Apple Podcasts and Spotify.
00:01:56.159 --> 00:01:58.400
And then go to your everydayai.com.
00:01:58.480 --> 00:02:07.040
We're going to be recapping all of the highlights from today's show as well as all of the other AI developments you need to know to make good decisions for your company.
00:02:07.200 --> 00:02:22.960
And yeah, there's going to be a lot new, I'm guessing, updates from today, uh, with the updates to the anthropic mythos release, then pullback, then re-release, and now we have GPT 5, 6 and three new variations.
00:02:23.199 --> 00:02:26.800
So uh let's start there, shall we?
00:02:26.879 --> 00:02:35.840
Uh, because we do have a new leading model from OpenAI, but unfortunately looks like hardly anyone has access to it.
00:02:35.919 --> 00:02:52.319
Uh, so OpenAI says it is limiting its early access to its newest GPT 5.6 models to a small group of trusted testers after requests from the US government raising new questions about how much control Washington should have over advanced AI releases.
00:02:52.560 --> 00:03:08.800
So OpenAI says the first preview of GPT 5.6 is restricted to partners whose participation has been shared with the US government rather than their normal uh kind of release cycle of just opening it up to millions of users right away.
00:03:09.120 --> 00:03:19.599
So aside from this kind of new-ish way to roll out models, which is just going to selected few, open AI also has a new naming scheme.
00:03:19.759 --> 00:03:20.800
So here we go.
00:03:20.879 --> 00:03:25.680
We have Soul, uh, which OpenAI calls its flagship and most powerful model.
00:03:25.840 --> 00:03:30.560
Then we have Terra, a more balanced version for everyday use.
00:03:30.800 --> 00:03:35.039
And then we have Luna, which is a faster, lower cost option option.
00:03:35.199 --> 00:03:43.120
So uh similar to the older anthropic three-tier naming capabilities, it looks like that's what we're gonna have from OpenAI moving on.
00:03:43.280 --> 00:03:49.680
So uh what that means is well, you have your sun, your earth, and your moon in that order.
00:03:49.759 --> 00:03:53.840
So Sol is the uh the big heavy uh model.
00:03:54.560 --> 00:04:00.960
Uh, then you have Terra, which is kind of the in-between or the old sonnet from Anthropic, if you're used to that naming mechanism.
00:04:01.120 --> 00:04:02.719
And then you have Luna, right?
00:04:02.800 --> 00:04:03.439
Our moon.
00:04:03.520 --> 00:04:05.919
Uh, that's the faster, lower cost option.
00:04:06.000 --> 00:04:09.919
So, yeah, think of your globe, uh, you know, sun, earth, moon.
00:04:10.000 --> 00:04:12.319
Yeah, we're gonna have to start memorizing these things.
00:04:12.479 --> 00:04:26.399
So OpenAI says also that Seoul adds a max reasoning mode and an ultra mode that uses coordinated subagents for very complex tasks, especially in coding biology and cybersecurity.
00:04:26.560 --> 00:04:37.360
So the company says GPT 5.6 soul is slightly better at coding workflows than Anthropics Clawed Mythos 5, while using about one-third of the output tokens.
00:04:37.439 --> 00:04:39.360
That is the big news there.
00:04:39.600 --> 00:04:53.519
So uh we don't have full benchmarks on the new GPT-5.6 soul, but it looks like from what OpenAI shared, it does look like it's in the mythos and fable range on some benchmarks.
00:04:53.680 --> 00:04:56.160
On some benchmarks, uh Anthropic is ahead.
00:04:56.240 --> 00:05:04.399
So we'll see when it comes to third party uh analysis like artificial analysis or uh the LM or formerly LM arena.
00:05:04.480 --> 00:05:11.839
So we'll see where it clocks in, but it looks like it's probably a little bit below uh kind of true mythos uh level.
00:05:11.920 --> 00:05:19.519
So maybe it's more in the fable five uh area, which is mythos without all or with that is mythos uh with guardrails.
00:05:19.680 --> 00:05:21.600
So yeah, it's super confusing now, right?
00:05:22.000 --> 00:05:29.759
Because uh three weeks ago we didn't have mythos and fable and soul and terra and luna, right?
00:05:29.839 --> 00:05:42.639
So now we have not only this new tier of maybe AI not being truly democratized anymore, and it's gonna be, you know, who knows if the general public will ever get uh access to mythos.
00:05:42.959 --> 00:05:55.920
Maybe it'll just be a limited availability with uh Fable, and seems like there's some rumors circulating that you know, in the future you might actually have to show US citizenship to use some of these uh models.
00:05:56.000 --> 00:05:58.319
So we'll see where the US government lands with that.
00:05:58.399 --> 00:06:00.319
But it is going to be one of the hottest topics.
00:06:00.480 --> 00:06:05.920
But uh regardless, it looks like a very promising release from OpenAI.
00:06:06.639 --> 00:06:12.000
So the Trump administration has tightened oversight of Frontier AI releases.
00:06:12.160 --> 00:06:26.319
And Dean Ball, a former White House AI advisor and now recently hired by OpenAI, says a recent executive order has effectively created an involuntary licensing system for most advanced models.
00:06:26.480 --> 00:06:40.319
So Ball argues that if the government does not define clear safety guardrails, AI launches could face repeated delays, which could slow US competition with China and complicate larger infrastructure investments.
00:06:40.480 --> 00:06:53.519
Uh, so OpenAI says its new models include stronger built-in security with defenses meant to resist jailbreaks and to prioritize defensive cybersecurity to help offensive misuse.
00:06:53.839 --> 00:07:06.720
OpenAI also says it is moving away from a separate filter system and toward safety guardrails built in the model itself after backlash over Anthropic's earlier approach of routing risky prompts to older models.
00:07:06.959 --> 00:07:09.199
So, yeah, we'll see, right?
00:07:09.519 --> 00:07:14.000
OpenAI said they hope to roll out access to the new GPT-5.
00:07:15.120 --> 00:07:17.839
Uh, to everyone in the coming weeks.
00:07:18.000 --> 00:07:32.879
Uh, we'll see where we land with that because uh Anthropic, as an example, uh, when they had to pull Fable 5 access, which they released to all paid users, they said that they would have some sort of communication update in 24 hours.
00:07:33.040 --> 00:07:37.360
It's been more than two weeks, and we don't have an official word from Anthropic.
00:07:37.439 --> 00:07:43.680
Uh, so open AI kind of promising the GPT-5-6 rollout to more people uh in the coming weeks.
00:07:43.759 --> 00:08:08.160
They were pretty um pretty forthright in their announcement, talking about some of these complications uh of rolling this model out and kind of this new reality we live in now, apparently, which is uh AI is no longer, at least temporarily, maybe permanently, truly democratized, and there may be a tiered system in terms of who has access to what level of intelligence.
00:08:08.319 --> 00:08:10.800
All right, sticking with a similar story.
00:08:10.959 --> 00:08:17.519
Well, Mythos is starting to be re-rolled out, but only to trusted users.
00:08:18.000 --> 00:08:20.560
So, uh, yeah, a lot of action on Friday.
00:08:20.639 --> 00:08:29.360
So this was, I believe, shortly after uh OpenAI announced GPT-5.6 and its limited release to trusted testers.
00:08:29.519 --> 00:08:45.679
So now the US government on Friday uh apparently is allowing Anthropic to re release its Mythos 5 models to only about 100 companies and a federal agency, marking a key shift after a two-week standoff with the Trump administration.
00:08:45.840 --> 00:09:12.799
So the Commerce Department, according to reports, said in a letter that appropriate safeguards are in place for certain trusted partners to access the Claude Mythos 5 models, which suggests the government is still limiting broad access to Mythos V and well, Fable 5, because the approval does not cover Anthropic's Fable 5 model, meaning the company still does not have permission to restore access to its newest, most advanced release.
00:09:12.960 --> 00:09:17.679
But you know, obviously that could change any hour, any minute.
00:09:17.840 --> 00:09:20.799
So, you know, make sure that you check today's newsletter.
00:09:20.960 --> 00:09:25.120
It's funny because actually on Friday, you know, we always do A B tests in the newsletter.
00:09:25.200 --> 00:09:33.360
So I sent the A B test, and then literally three minutes later, uh, we got the uh the release or the news from OpenAI.
00:09:33.600 --> 00:09:40.399
So uh, you know, we'll see if there is any uh new update, uh at least when it comes to Anthropic.
00:09:40.480 --> 00:09:56.879
But Anthropic has disabled access to both Fable V and Mythos V earlier this month after an exports control directive cited national security concerns and ordered access suspended for four nationals, which included four national employees.
00:09:57.120 --> 00:10:07.919
So the move came only days after Anthropic unveiled the models and said Fable 5 was its first public release of such an advanced system backed by new safeguards.
00:10:08.000 --> 00:10:11.679
The company said blocked responses in high-risk areas.
00:10:11.840 --> 00:10:17.600
So essentially, Anthropic after that directive from the US government said, Well, there's no way to actually do that.
00:10:17.679 --> 00:10:23.279
So they had to pull the model worldwide, and that's where we still stand, aside from this new update.
00:10:23.360 --> 00:10:32.639
Of apparently the US government is letting those uh roughly 100 companies and federal agencies get access back to Mythos.
00:10:33.679 --> 00:10:47.759
So Anthropics negotiations with the Trump administration have also reportedly shifted from CEO Dario Mati to co-founder Tom Brown Brown, uh, underscoring how serious and sensitive the talks have become.
00:10:47.919 --> 00:10:51.200
So, yeah, this one was kind of interesting to see.
00:10:51.360 --> 00:11:02.639
According uh to reports, there were some White House officials uh that uh alluded that Dario was being a weirdo and they couldn't really get talks going with him.
00:11:02.799 --> 00:11:09.360
So instead, Anthropic co-founder uh Tom Brown uh was tapped in foreign negotiations.
00:11:09.440 --> 00:11:18.080
And it seems like, at least from Anthropic's point of view, uh things are going better than they were, at least last week or for the rest of June.
00:11:18.240 --> 00:11:28.799
Uh, the company has also faced pressure from the defense department, which previously labeled Anthropic a supply chain risk, a rare designation, usually reserved for foreign adversaries.
00:11:29.039 --> 00:11:40.960
Anthropic has sued the Trump administration over that original uh blacklisting, and the legal fight is still ongoing, adding another layer of uncertainty around the company's government business.
00:11:41.600 --> 00:11:48.399
All right, so back to the other uh AI company as part of the big four, right?
00:11:48.480 --> 00:11:51.759
We have open AI, Anthropic, Google, and Microsoft.
00:11:51.840 --> 00:11:57.360
So we just talked about where uh OpenAI and Anthropic stand with their latest models.
00:11:57.519 --> 00:11:58.639
Now, what about Google?
00:11:58.720 --> 00:12:11.759
Because at last month's Google IO conference, when they announced and released their Gemini 3.5 flash, they said that their big model, Gemini 3.5 Pro, would be unveiled in June.
00:12:11.840 --> 00:12:17.200
And here we are in the last hours of June, and we don't have any new model from Google yet.
00:12:17.279 --> 00:12:30.559
So, according to reports, it is kind of the uh pressure now from the US government to get this model approved, but there's also allegedly some other things going on behind the curtains at Google.
00:12:30.720 --> 00:12:48.000
So, according to reporting from the information and other outlets, Google is turning its short-term AI coding strike team into a permanent mid-training group to help close the gap with anthropic and AI assistant coding development.
00:12:48.399 --> 00:13:04.559
So, uh, this move matters because Google appears to be reshaping how it trains Gemini, shifting from relying mostly on broad pre-training and fine-tuning to adding a formal middle training stage with specialized coding data.
00:13:04.720 --> 00:13:14.879
So that stage is called mid-training, and it's meant to teach the model more than code syntax, including how to work through larger engineering tasks, tool use, and multi-file code understanding.
00:13:15.039 --> 00:13:32.159
So the reorganization follows pressure from Anthropic's Clawed Code, which has become a major competitor in AI coding, where internal estimates reportedly put Google engineers at about 50% AI assistance versus nearly a hundred percent reportedly at anthropic.
00:13:32.320 --> 00:13:50.720
The timing is also tied to key talent loss, with several notable researchers in engineering and engineers leaving Google's four rivals, including uh Jonas Adler, uh Alexander Pritzel, John Jumper, and Noam Shazier, according to reporting.
00:13:50.879 --> 00:14:02.159
So, yeah, Google has kind of lost an all-star team of AI engineers over the past few weeks, uh, which led to them reportedly turning this strike team, right?
00:14:02.240 --> 00:14:06.000
So it's kind of a different variation of these code reds.
00:14:06.080 --> 00:14:11.600
You hear these companies, you know, uh push out internally when it's like, oh, it seems like we're kind of behind.
00:14:11.679 --> 00:14:13.039
We better call a code red.
00:14:13.360 --> 00:14:16.960
So that's kind of the equivalent of what this strike team is, right?
00:14:17.039 --> 00:14:24.559
They said, hey, a couple of months ago, according to reports, that they realized internally that they were falling behind on coding.
00:14:24.639 --> 00:14:29.039
So they developed what was supposed to be a temporary coding strike team.
00:14:29.120 --> 00:14:37.679
And now apparently, uh, that uh you know, kind of uh realization that maybe they're a little further behind uh on coding.
00:14:37.840 --> 00:14:42.080
This is according to reports, not according to me or my views, uh, right.
00:14:42.159 --> 00:14:45.279
So that combined with losing some major talent.
00:14:45.360 --> 00:14:48.639
I mean, those are some names if you follow AI.
00:14:48.799 --> 00:14:51.759
I mean, those are some of the OGs of AI.
00:14:51.919 --> 00:14:59.679
So uh Google originally formed that team in April um to speed up Gemini's coding abilities.
00:14:59.840 --> 00:15:02.320
So yeah, it could be a long-term thing.
00:15:02.399 --> 00:15:14.559
And the updated roadmap now goes beyond simple code completion and is aimed at long-running engineering agents that can carry out multi-step tasks over time with less human supervision.
00:15:14.720 --> 00:15:30.000
So, yeah, we'll see uh what happens, and we'll be continuing to follow uh all these developments from Google because yeah, they did publicly say in May that Gemini 3.5 Pro would be out next month, which is June, right?
00:15:30.080 --> 00:15:31.840
And how many days do we have left of June?
00:15:31.919 --> 00:15:36.159
We have uh, I don't know, about 30 some hours left of June.
00:15:36.240 --> 00:15:39.360
So we'll see if we get an official announcement from Google.
00:15:39.519 --> 00:15:45.679
I would expect no, we won't just because of this new climate that we are now living in.
00:15:46.000 --> 00:16:03.919
All right, our next piece of AI news a group of AI uh juggernauts and former government workers have combined to create a new nonprofit helped at well raising up the employability of workers.
00:16:04.159 --> 00:16:18.159
So uh a new nonprofit called Raise US says it has already raised$500 million toward a$1 billion goal to help states in the US prepare for AI-related disruption.
00:16:18.320 --> 00:16:32.639
So the group is being led by former commerce secretary, uh Gina uh Gina Ramondo, who said the US has a tech strategy for AI competition, but still lacks a clear people strategy for workers.
00:16:32.799 --> 00:16:43.759
So Ray's US is based by major players in tech and AI, including the Open AI Foundation, Anthropic, Amazon, and Microsoft, which are serving as anchor partners.
00:16:43.919 --> 00:16:55.919
So the nonprofit says it will build a national platform to advise state governors on how to prepare local workforces for AI-driven change with a focus on practical policy and training models.
00:16:56.240 --> 00:17:00.480
So far, uh there are some notable state partnerships already, right?
00:17:00.559 --> 00:17:03.200
Kind of hours after we got win of this new group.
00:17:03.360 --> 00:17:15.039
So the first state partnerships are all already underway with Arkansas, Connecticut, Maryland, and Utah split evenly between Republican and Democratic-led states.
00:17:15.200 --> 00:17:29.279
In Arkansas, Ray's US is working with governor uh Governor Sarah Huckabee Sanders on Arkansas launch, an AI-powered career navigation platform that would connect students and job seekers to personalized learning and employer-linked career paths.
00:17:29.440 --> 00:17:38.400
In Maryland, the group is working with Governor Wes Moore to expand its service year opportunities for recent high school graduates in fields like healthcare and education.
00:17:38.720 --> 00:17:51.359
Utah is a notable test because the state has been forced into the national spotlight over the build-out of AI data centers, showing how AI expansion can create both job and infrastructure tensions.
00:17:51.519 --> 00:18:05.359
So RAISE US says more states are expected to join in the coming months and also wants to test policies such as short time compensation and wage insurance that could soften the blow of job changes.
00:18:05.519 --> 00:18:11.599
So the effort comes as debate continues over how much AI will affect the labor market.
00:18:12.000 --> 00:18:31.599
As we know, we heard uh kind of some uh wild uh kind of predictions, such as that from Anthropic CEO Dario Mati, who originally said that AI would disrupt up to 50% of white-collar jobs, which obviously didn't happen, and he kind of had to adjust or update his stance on that.
00:18:31.680 --> 00:18:39.200
But regardless, we are already seeing in 2026 an unprecedented uh impact of AI on full-time jobs.
00:18:39.440 --> 00:18:58.240
So uh it's good to see a company come in and address the human side, which I've always said since I started this podcast three and a half years ago that AI implementation is much more about people and man, uh people management, change management than it is about you know traditional tech implementation or digital transformation.
00:18:58.400 --> 00:19:10.079
So, all right, our next piece of AI news we have a fresh accusation of Anthropic pointing their fingers at the Chinese saying you stole our models.
00:19:10.240 --> 00:19:25.759
So, according to C NBC, Anthropic sent a letter this month to the Senate banking committee accusing Alibaba of carrying out what it called the largest known distillation attack on Anthropic's AI models.
00:19:25.920 --> 00:19:42.240
So Anthropic said the activity involved about 28 million exchanges through roughly 25,000 fraudulent accounts between April and June, which it says points to a large-scale attempt to copy its model behavior.
00:19:42.400 --> 00:19:55.440
So the company told lawmakers that operators linked to Alibaba and its AI lab were behind the effort, framing the issue as a major example of AI model theft rather than normal competition.
00:19:55.519 --> 00:20:02.400
So distillation is a process, if you don't know, where a smaller model learns from the outputs of a stronger one.
00:20:02.559 --> 00:20:13.359
But Anthropic is accusing the Alibaba-linked effort of using that method to, well, just steal all of their work and infuse it into Alibaba's models.
00:20:13.599 --> 00:20:23.039
Anthropic said fighting this kind of AI copying will require coordination between government and industry, and that it wants Washington to help protect U.S.
00:20:23.200 --> 00:20:24.400
leadership.
00:20:24.720 --> 00:20:39.519
So, according to reports, the letter comes two months after the White House Office of Science and Technology Policy said it would help AI companies in the US detect and respond to industrial-scale distillation attempts.
00:20:39.680 --> 00:20:50.960
Anthropic said Alibaba's actions ignored the Trump administration's warnings, uh, making the disputes part of the bigger US-China technology and national security fight.
00:20:51.200 --> 00:21:00.720
So this is definitely a big story to keep a watch on because, well, this also impacts what models your company may or may not be using.
00:21:00.880 --> 00:21:06.720
So last week, make sure you go check out uh our start here series if you haven't already.
00:21:06.960 --> 00:21:09.599
But in episode uh which one was that?
00:21:09.759 --> 00:21:27.599
In episode 29 of the Start Here series, which was episode 804 of Everyday AI, we talked about a new open source model called GLM52, which may be the first time we have a Chinese open weight or open source model that is almost at the frontier level.
00:21:27.839 --> 00:21:29.839
So it's gonna be pretty interesting.
00:21:29.920 --> 00:21:41.119
And I think this uh probably deserves a larger conversation if the access to true frontier models is no longer actually democratized, right?
00:21:41.200 --> 00:21:56.720
If the 99.9% of US workers, at least right now, right, can't access Anthropics Fable, they can't access uh Anthropics mythos, they can't access right now, at least maybe that might change in a week or two.
00:21:56.880 --> 00:21:59.279
They can't access, you know, GPT 56 soul.
00:22:00.319 --> 00:22:04.880
So when you have these other Chinese models that are starting to get in that tier, right?
00:22:04.960 --> 00:22:10.160
I wouldn't say they're in the tier of GPT 5.6 and Mythos, but they're in the tier right below, right?
00:22:10.240 --> 00:22:14.240
Which is, well, models available to every single person, right?
00:22:14.400 --> 00:22:18.720
So that's GPT-5.5 and Opus 4.8 from Anthropic.
00:22:18.960 --> 00:22:20.960
So as you get these Chinese models, right?
00:22:21.119 --> 00:22:31.039
Whether they're from Alibaba, uh, they're from ZAI, whoever they are, that are starting to get in that tier of the best publicly available models, it is an important conversation to keep an eye on.
00:22:31.119 --> 00:22:50.000
So distillation, it's it's the combination of distillation and maybe these frontier systems only being available to the select few uh creates for a very interesting and mixy uh uh sorry, a potentially um explosive combination, right?
00:22:50.160 --> 00:22:58.559
In terms of what models your company should use and well, what it means for revenue of companies like anthropic who are preparing to go public.
00:22:59.279 --> 00:22:59.759
All right.
00:23:00.079 --> 00:23:04.319
Our last big piece of AI news this week is chip news.
00:23:04.480 --> 00:23:18.079
Yeah, because open AI and Broadcom have unveiled Jalapeno, a custom chip meant to run large language models more efficiently, with deployment expected to happen later this year.
00:23:18.240 --> 00:23:44.319
So the big picture goal for OpenAI and Broadcom was to make an AI systems uh to make their AI systems cheaper and faster and less power hungry to run at scale, which matters a lot because energy and computing costs are becoming major limits for AI companies and their ability to provide uh frontier AI models to the broader enterprise um lens.
00:23:44.880 --> 00:23:59.680
So, with this new chip partnership, OpenAI's first, open AI says jalapeno is built specifically for AI inference, meaning it is designed to help models answer questions and serve users rather than train them, right?
00:23:59.839 --> 00:24:06.079
So, obviously, on the training side, Nvidia's GPUs have been the industry standard for a long time.
00:24:06.400 --> 00:24:17.359
This chip is specifically built for inference, like all of the back and forth that open AI needs when uh up to a billion users each week are using their services.
00:24:17.599 --> 00:24:27.119
So the companies say the chip is aimed at the hardest parts of the AI service, including speed, memory use, networking, and reducing wasted compute.
00:24:27.279 --> 00:24:37.279
So OpenAI and Broadcom claim the chip will be more efficient than current leading hardware, but they have not shared benchmarks or hard performance numbers yet.
00:24:37.519 --> 00:24:46.079
Images shared by the company suggest a very large chip with multiple memory modules, which points to a serious effort to handle heavy AI workloads.
00:24:46.240 --> 00:24:52.400
So the companies say early samples are already working in the lab, which is a crazy fast turnaround.
00:24:52.480 --> 00:24:55.680
And OpenAI says it is testing real workloads on the chips now.
00:24:55.920 --> 00:25:10.000
OpenAI and Broadcom also say the chip was developed in just nine months, which is much faster than typical uh typical chip development and shows how quickly AI infrastructure is being pushed forward.
00:25:10.240 --> 00:25:15.519
All right, so that's a wrap for our major big stories of the week.
00:25:16.319 --> 00:25:20.319
But let's quickly move into what's new and what's next.
00:25:20.480 --> 00:25:28.559
So this is a bullet point roundup of everything else noteworthy that either happened last week or is happening this week.
00:25:28.720 --> 00:25:34.160
So it's a mixture of smaller news stories, maybe some leaks and rumors, but let's get straight into it.
00:25:34.319 --> 00:25:35.680
Fast bullet point style.
00:25:35.759 --> 00:25:36.160
Here we go.
00:25:36.319 --> 00:25:41.920
So nearly 400 newspapers have sued OpenAI in Microsoft over content.
00:25:42.160 --> 00:25:48.240
Qualcomm and Meta announced a multi-generation data center uh CPU partnership.
00:25:48.400 --> 00:25:54.000
Uh Chevron signed a 20-year Microsoft power deal for AI data centers.
00:25:54.079 --> 00:26:05.920
Uh, Meta launched a$299 pair of AI glasses uh in line with SLR Luxotica, the famous ibrand maker that I probably did not pronounce correctly.
00:26:06.000 --> 00:26:11.839
Uh Google reportedly limited Meta's access to Gemini AI models over capacity shortages.
00:26:12.079 --> 00:26:21.200
Elon Musk's says that XAI's new 1.5 trillion parameter Groc 4.5 has begun private beta testing.
00:26:21.279 --> 00:26:28.720
So we'll see if they have to go through the normal uh government uh hoops uh jumping through those or if they'll be able to release it.
00:26:28.960 --> 00:26:34.079
Uh Micron and Anthropic sign a memory deal to help scale Claude.
00:26:34.400 --> 00:26:42.160
Apple's VP4 Vision Pro and smart glasses is leaving for OpenAI's hardware unit, according to reports.
00:26:42.319 --> 00:26:50.319
Uh leaks show that OpenAI is testing computer use with Excel and PowerPoint add-ins in codecs, which will be an interesting update.
00:26:50.480 --> 00:26:57.039
Uh, Google, inside of Notebook LM, added fully customizable flashcards for users to edit and share.
00:26:57.200 --> 00:27:07.440
OpenAI reportedly is weighing a 2027 IPO instead of 2026 after some recent SpaceX volatility.
00:27:07.599 --> 00:27:16.480
In an internal study, OpenAI said Codecs accounted for 99.8% of internal weekly AI output tokens, which is a wild percentage.
00:27:16.640 --> 00:27:22.720
Almost 100% of their output tokens internally are in codecs, not in their chat GPT service.
00:27:22.960 --> 00:27:29.039
OpenAI released daybreak cyber tools, including GPT-5.5 cyber and codec security updates.
00:27:29.200 --> 00:27:35.920
OpenAI released a refreshed GPT 5.5 instant model with better conversation and handling and constraints.
00:27:36.160 --> 00:27:42.559
Google launched Google Finance in general availability with portfolio tracking and AI research tools.
00:27:42.720 --> 00:27:47.519
Google released Gemini for home smart speakers for natural voice conversations.
00:27:47.759 --> 00:27:55.759
Anthropic launched Claude Tag in Slack, which is kind of like a manager per Slack channel that can do anything that uh Claude can do.
00:27:56.000 --> 00:28:01.119
Canva released Canva Grow 2.0 for AI ad creation, publishing, and optimization.
00:28:01.359 --> 00:28:12.240
Google added Select from Screen in Chrome's Gemini sidebar after they updated their Gemini 3.5 uh flash to have computer use capabilities.
00:28:12.400 --> 00:28:24.079
Microsoft added Copilot skills in Excel for reputable task automation and a new digital ocean plugin for codecs, entered public preview for remote cloud dev machines.
00:28:24.240 --> 00:28:25.519
All right, that's a lot.
00:28:25.599 --> 00:28:31.279
And I can almost guarantee you this week is going to move even faster than last week.
00:28:31.359 --> 00:28:36.000
Because y'all, this everything with OpenAI and Anthropic in terms of GBD5.
00:28:39.039 --> 00:28:47.759
The government uh kind of asking them to hold it back, the re-release, the slow re-release of Anthropic Mythos, all of this just happened Friday.
00:28:47.839 --> 00:28:52.000
So I expect a ton of new developments that you need to be paying attention to.
00:28:52.319 --> 00:29:00.400
So don't spend hours every single day sitting around your boardroom talking about AI strategy and looking at that day's news, right?
00:29:00.559 --> 00:29:02.000
Just let me do that for you.
00:29:02.160 --> 00:29:05.440
Then join us on Mondays as we bring you the AI news that matters.
00:29:05.599 --> 00:29:08.000
If you're new here, on Wednesdays, we go hands-on.
00:29:08.160 --> 00:29:09.359
We usually do a demo.
00:29:09.440 --> 00:29:14.799
Uh, so make sure you check out today's newsletter so you can let us know what you want to see on that show.
00:29:14.960 --> 00:29:19.759
On Fridays, we go over fresh uh AI features that are available to everyone.
00:29:19.920 --> 00:29:23.119
And then on Tuesdays and Thursdays, we rotate the type of show.
00:29:23.279 --> 00:29:28.000
So uh please, if you haven't already, subscribe to the show on Spotify and Apple Podcasts.
00:29:28.079 --> 00:29:32.319
Then go to your everydayai.com, sign up for the free daily newsletter.
00:29:32.400 --> 00:29:33.119
Thanks for tuning in.
00:29:33.200 --> 00:29:36.160
We hope to see you back tomorrow and every day for more everyday AI.
00:29:36.240 --> 00:29:37.279
Thanks, y'all.
00:00:01.919 --> 00:00:06.480
This June was supposed to be the start of Hot AI summer.
00:00:06.719 --> 00:00:19.519
That's because earlier this month, Anthropic released its Mythos 5 and Fable 5 models to the general public, and we were supposed to have an avalanche of other Frontier models released to all.
00:00:19.679 --> 00:00:28.000
Google was slated to release Gemini 3.5 Pro, and OpenAI was on board to drop GPT 5.6 all this month.
00:00:28.160 --> 00:00:40.079
And although we technically did kind of get the GPT-5 release from OpenAI late Friday, it's not exactly the hot AI summer we were promised.
00:00:40.240 --> 00:00:46.880
That's because Anthropic's heavy capabilities hypings of Mythos caused the US government to step in.
00:00:47.119 --> 00:00:55.920
And now we're at the end of June with no general availability to the promised frontier models from any of the big three companies.
00:00:56.079 --> 00:01:08.799
Although we do have some updated news because Mythos 5 and the new GPT 5.6 have started to roll out, but only if you work at a select few companies.
00:01:08.959 --> 00:01:12.079
So that's the big AI news of the week.
00:01:12.159 --> 00:01:15.040
But there's a lot more that we're going to get you caught up on.
00:01:15.200 --> 00:01:23.200
So if you don't have hours every single day trying to understand what's happening in the AI world and how it impacts your business, don't worry.
00:01:23.359 --> 00:01:25.120
That's what I do every single day.
00:01:25.200 --> 00:01:28.159
And then on Monday, we bring you the AI news that matters.
00:01:28.400 --> 00:01:29.760
So let's get into it.
00:01:29.840 --> 00:01:30.879
If you're new here, welcome.
00:01:30.959 --> 00:01:33.599
My name is Jordan Wilson, and this is Everyday AI.
00:01:33.760 --> 00:01:46.480
It's a daily live stream podcast and free daily news that are helping everyday business leaders like you and me not just keep up, but how we can use all of this information hidden between the headlines to get ahead and grow our companies and careers.
00:01:46.560 --> 00:01:50.319
So it starts here with the unedited, unscripted live stream podcast.
00:01:50.400 --> 00:01:56.079
But please make sure if you haven't already, subscribe to the podcast on Apple Podcasts and Spotify.
00:01:56.159 --> 00:01:58.400
And then go to your everydayai.com.
00:01:58.480 --> 00:02:07.040
We're going to be recapping all of the highlights from today's show as well as all of the other AI developments you need to know to make good decisions for your company.
00:02:07.200 --> 00:02:22.960
And yeah, there's going to be a lot new, I'm guessing, updates from today, uh, with the updates to the anthropic mythos release, then pullback, then re-release, and now we have GPT 5, 6 and three new variations.
00:02:23.199 --> 00:02:26.800
So uh let's start there, shall we?
00:02:26.879 --> 00:02:35.840
Uh, because we do have a new leading model from OpenAI, but unfortunately looks like hardly anyone has access to it.
00:02:35.919 --> 00:02:52.319
Uh, so OpenAI says it is limiting its early access to its newest GPT 5.6 models to a small group of trusted testers after requests from the US government raising new questions about how much control Washington should have over advanced AI releases.
00:02:52.560 --> 00:03:08.800
So OpenAI says the first preview of GPT 5.6 is restricted to partners whose participation has been shared with the US government rather than their normal uh kind of release cycle of just opening it up to millions of users right away.
00:03:09.120 --> 00:03:19.599
So aside from this kind of new-ish way to roll out models, which is just going to selected few, open AI also has a new naming scheme.
00:03:19.759 --> 00:03:20.800
So here we go.
00:03:20.879 --> 00:03:25.680
We have Soul, uh, which OpenAI calls its flagship and most powerful model.
00:03:25.840 --> 00:03:30.560
Then we have Terra, a more balanced version for everyday use.
00:03:30.800 --> 00:03:35.039
And then we have Luna, which is a faster, lower cost option option.
00:03:35.199 --> 00:03:43.120
So uh similar to the older anthropic three-tier naming capabilities, it looks like that's what we're gonna have from OpenAI moving on.
00:03:43.280 --> 00:03:49.680
So uh what that means is well, you have your sun, your earth, and your moon in that order.
00:03:49.759 --> 00:03:53.840
So Sol is the uh the big heavy uh model.
00:03:54.560 --> 00:04:00.960
Uh, then you have Terra, which is kind of the in-between or the old sonnet from Anthropic, if you're used to that naming mechanism.
00:04:01.120 --> 00:04:02.719
And then you have Luna, right?
00:04:02.800 --> 00:04:03.439
Our moon.
00:04:03.520 --> 00:04:05.919
Uh, that's the faster, lower cost option.
00:04:06.000 --> 00:04:09.919
So, yeah, think of your globe, uh, you know, sun, earth, moon.
00:04:10.000 --> 00:04:12.319
Yeah, we're gonna have to start memorizing these things.
00:04:12.479 --> 00:04:26.399
So OpenAI says also that Seoul adds a max reasoning mode and an ultra mode that uses coordinated subagents for very complex tasks, especially in coding biology and cybersecurity.
00:04:26.560 --> 00:04:37.360
So the company says GPT 5.6 soul is slightly better at coding workflows than Anthropics Clawed Mythos 5, while using about one-third of the output tokens.
00:04:37.439 --> 00:04:39.360
That is the big news there.
00:04:39.600 --> 00:04:53.519
So uh we don't have full benchmarks on the new GPT-5.6 soul, but it looks like from what OpenAI shared, it does look like it's in the mythos and fable range on some benchmarks.
00:04:53.680 --> 00:04:56.160
On some benchmarks, uh Anthropic is ahead.
00:04:56.240 --> 00:05:04.399
So we'll see when it comes to third party uh analysis like artificial analysis or uh the LM or formerly LM arena.
00:05:04.480 --> 00:05:11.839
So we'll see where it clocks in, but it looks like it's probably a little bit below uh kind of true mythos uh level.
00:05:11.920 --> 00:05:19.519
So maybe it's more in the fable five uh area, which is mythos without all or with that is mythos uh with guardrails.
00:05:19.680 --> 00:05:21.600
So yeah, it's super confusing now, right?
00:05:22.000 --> 00:05:29.759
Because uh three weeks ago we didn't have mythos and fable and soul and terra and luna, right?
00:05:29.839 --> 00:05:42.639
So now we have not only this new tier of maybe AI not being truly democratized anymore, and it's gonna be, you know, who knows if the general public will ever get uh access to mythos.
00:05:42.959 --> 00:05:55.920
Maybe it'll just be a limited availability with uh Fable, and seems like there's some rumors circulating that you know, in the future you might actually have to show US citizenship to use some of these uh models.
00:05:56.000 --> 00:05:58.319
So we'll see where the US government lands with that.
00:05:58.399 --> 00:06:00.319
But it is going to be one of the hottest topics.
00:06:00.480 --> 00:06:05.920
But uh regardless, it looks like a very promising release from OpenAI.
00:06:06.639 --> 00:06:12.000
So the Trump administration has tightened oversight of Frontier AI releases.
00:06:12.160 --> 00:06:26.319
And Dean Ball, a former White House AI advisor and now recently hired by OpenAI, says a recent executive order has effectively created an involuntary licensing system for most advanced models.
00:06:26.480 --> 00:06:40.319
So Ball argues that if the government does not define clear safety guardrails, AI launches could face repeated delays, which could slow US competition with China and complicate larger infrastructure investments.
00:06:40.480 --> 00:06:53.519
Uh, so OpenAI says its new models include stronger built-in security with defenses meant to resist jailbreaks and to prioritize defensive cybersecurity to help offensive misuse.
00:06:53.839 --> 00:07:06.720
OpenAI also says it is moving away from a separate filter system and toward safety guardrails built in the model itself after backlash over Anthropic's earlier approach of routing risky prompts to older models.
00:07:06.959 --> 00:07:09.199
So, yeah, we'll see, right?
00:07:09.519 --> 00:07:14.000
OpenAI said they hope to roll out access to the new GPT-5.
00:07:15.120 --> 00:07:17.839
Uh, to everyone in the coming weeks.
00:07:18.000 --> 00:07:32.879
Uh, we'll see where we land with that because uh Anthropic, as an example, uh, when they had to pull Fable 5 access, which they released to all paid users, they said that they would have some sort of communication update in 24 hours.
00:07:33.040 --> 00:07:37.360
It's been more than two weeks, and we don't have an official word from Anthropic.
00:07:37.439 --> 00:07:43.680
Uh, so open AI kind of promising the GPT-5-6 rollout to more people uh in the coming weeks.
00:07:43.759 --> 00:08:08.160
They were pretty um pretty forthright in their announcement, talking about some of these complications uh of rolling this model out and kind of this new reality we live in now, apparently, which is uh AI is no longer, at least temporarily, maybe permanently, truly democratized, and there may be a tiered system in terms of who has access to what level of intelligence.
00:08:08.319 --> 00:08:10.800
All right, sticking with a similar story.
00:08:10.959 --> 00:08:17.519
Well, Mythos is starting to be re-rolled out, but only to trusted users.
00:08:18.000 --> 00:08:20.560
So, uh, yeah, a lot of action on Friday.
00:08:20.639 --> 00:08:29.360
So this was, I believe, shortly after uh OpenAI announced GPT-5.6 and its limited release to trusted testers.
00:08:29.519 --> 00:08:45.679
So now the US government on Friday uh apparently is allowing Anthropic to re release its Mythos 5 models to only about 100 companies and a federal agency, marking a key shift after a two-week standoff with the Trump administration.
00:08:45.840 --> 00:09:12.799
So the Commerce Department, according to reports, said in a letter that appropriate safeguards are in place for certain trusted partners to access the Claude Mythos 5 models, which suggests the government is still limiting broad access to Mythos V and well, Fable 5, because the approval does not cover Anthropic's Fable 5 model, meaning the company still does not have permission to restore access to its newest, most advanced release.
00:09:12.960 --> 00:09:17.679
But you know, obviously that could change any hour, any minute.
00:09:17.840 --> 00:09:20.799
So, you know, make sure that you check today's newsletter.
00:09:20.960 --> 00:09:25.120
It's funny because actually on Friday, you know, we always do A B tests in the newsletter.
00:09:25.200 --> 00:09:33.360
So I sent the A B test, and then literally three minutes later, uh, we got the uh the release or the news from OpenAI.
00:09:33.600 --> 00:09:40.399
So uh, you know, we'll see if there is any uh new update, uh at least when it comes to Anthropic.
00:09:40.480 --> 00:09:56.879
But Anthropic has disabled access to both Fable V and Mythos V earlier this month after an exports control directive cited national security concerns and ordered access suspended for four nationals, which included four national employees.
00:09:57.120 --> 00:10:07.919
So the move came only days after Anthropic unveiled the models and said Fable 5 was its first public release of such an advanced system backed by new safeguards.
00:10:08.000 --> 00:10:11.679
The company said blocked responses in high-risk areas.
00:10:11.840 --> 00:10:17.600
So essentially, Anthropic after that directive from the US government said, Well, there's no way to actually do that.
00:10:17.679 --> 00:10:23.279
So they had to pull the model worldwide, and that's where we still stand, aside from this new update.
00:10:23.360 --> 00:10:32.639
Of apparently the US government is letting those uh roughly 100 companies and federal agencies get access back to Mythos.
00:10:33.679 --> 00:10:47.759
So Anthropics negotiations with the Trump administration have also reportedly shifted from CEO Dario Mati to co-founder Tom Brown Brown, uh, underscoring how serious and sensitive the talks have become.
00:10:47.919 --> 00:10:51.200
So, yeah, this one was kind of interesting to see.
00:10:51.360 --> 00:11:02.639
According uh to reports, there were some White House officials uh that uh alluded that Dario was being a weirdo and they couldn't really get talks going with him.
00:11:02.799 --> 00:11:09.360
So instead, Anthropic co-founder uh Tom Brown uh was tapped in foreign negotiations.
00:11:09.440 --> 00:11:18.080
And it seems like, at least from Anthropic's point of view, uh things are going better than they were, at least last week or for the rest of June.
00:11:18.240 --> 00:11:28.799
Uh, the company has also faced pressure from the defense department, which previously labeled Anthropic a supply chain risk, a rare designation, usually reserved for foreign adversaries.
00:11:29.039 --> 00:11:40.960
Anthropic has sued the Trump administration over that original uh blacklisting, and the legal fight is still ongoing, adding another layer of uncertainty around the company's government business.
00:11:41.600 --> 00:11:48.399
All right, so back to the other uh AI company as part of the big four, right?
00:11:48.480 --> 00:11:51.759
We have open AI, Anthropic, Google, and Microsoft.
00:11:51.840 --> 00:11:57.360
So we just talked about where uh OpenAI and Anthropic stand with their latest models.
00:11:57.519 --> 00:11:58.639
Now, what about Google?
00:11:58.720 --> 00:12:11.759
Because at last month's Google IO conference, when they announced and released their Gemini 3.5 flash, they said that their big model, Gemini 3.5 Pro, would be unveiled in June.
00:12:11.840 --> 00:12:17.200
And here we are in the last hours of June, and we don't have any new model from Google yet.
00:12:17.279 --> 00:12:30.559
So, according to reports, it is kind of the uh pressure now from the US government to get this model approved, but there's also allegedly some other things going on behind the curtains at Google.
00:12:30.720 --> 00:12:48.000
So, according to reporting from the information and other outlets, Google is turning its short-term AI coding strike team into a permanent mid-training group to help close the gap with anthropic and AI assistant coding development.
00:12:48.399 --> 00:13:04.559
So, uh, this move matters because Google appears to be reshaping how it trains Gemini, shifting from relying mostly on broad pre-training and fine-tuning to adding a formal middle training stage with specialized coding data.
00:13:04.720 --> 00:13:14.879
So that stage is called mid-training, and it's meant to teach the model more than code syntax, including how to work through larger engineering tasks, tool use, and multi-file code understanding.
00:13:15.039 --> 00:13:32.159
So the reorganization follows pressure from Anthropic's Clawed Code, which has become a major competitor in AI coding, where internal estimates reportedly put Google engineers at about 50% AI assistance versus nearly a hundred percent reportedly at anthropic.
00:13:32.320 --> 00:13:50.720
The timing is also tied to key talent loss, with several notable researchers in engineering and engineers leaving Google's four rivals, including uh Jonas Adler, uh Alexander Pritzel, John Jumper, and Noam Shazier, according to reporting.
00:13:50.879 --> 00:14:02.159
So, yeah, Google has kind of lost an all-star team of AI engineers over the past few weeks, uh, which led to them reportedly turning this strike team, right?
00:14:02.240 --> 00:14:06.000
So it's kind of a different variation of these code reds.
00:14:06.080 --> 00:14:11.600
You hear these companies, you know, uh push out internally when it's like, oh, it seems like we're kind of behind.
00:14:11.679 --> 00:14:13.039
We better call a code red.
00:14:13.360 --> 00:14:16.960
So that's kind of the equivalent of what this strike team is, right?
00:14:17.039 --> 00:14:24.559
They said, hey, a couple of months ago, according to reports, that they realized internally that they were falling behind on coding.
00:14:24.639 --> 00:14:29.039
So they developed what was supposed to be a temporary coding strike team.
00:14:29.120 --> 00:14:37.679
And now apparently, uh, that uh you know, kind of uh realization that maybe they're a little further behind uh on coding.
00:14:37.840 --> 00:14:42.080
This is according to reports, not according to me or my views, uh, right.
00:14:42.159 --> 00:14:45.279
So that combined with losing some major talent.
00:14:45.360 --> 00:14:48.639
I mean, those are some names if you follow AI.
00:14:48.799 --> 00:14:51.759
I mean, those are some of the OGs of AI.
00:14:51.919 --> 00:14:59.679
So uh Google originally formed that team in April um to speed up Gemini's coding abilities.
00:14:59.840 --> 00:15:02.320
So yeah, it could be a long-term thing.
00:15:02.399 --> 00:15:14.559
And the updated roadmap now goes beyond simple code completion and is aimed at long-running engineering agents that can carry out multi-step tasks over time with less human supervision.
00:15:14.720 --> 00:15:30.000
So, yeah, we'll see uh what happens, and we'll be continuing to follow uh all these developments from Google because yeah, they did publicly say in May that Gemini 3.5 Pro would be out next month, which is June, right?
00:15:30.080 --> 00:15:31.840
And how many days do we have left of June?
00:15:31.919 --> 00:15:36.159
We have uh, I don't know, about 30 some hours left of June.
00:15:36.240 --> 00:15:39.360
So we'll see if we get an official announcement from Google.
00:15:39.519 --> 00:15:45.679
I would expect no, we won't just because of this new climate that we are now living in.
00:15:46.000 --> 00:16:03.919
All right, our next piece of AI news a group of AI uh juggernauts and former government workers have combined to create a new nonprofit helped at well raising up the employability of workers.
00:16:04.159 --> 00:16:18.159
So uh a new nonprofit called Raise US says it has already raised$500 million toward a$1 billion goal to help states in the US prepare for AI-related disruption.
00:16:18.320 --> 00:16:32.639
So the group is being led by former commerce secretary, uh Gina uh Gina Ramondo, who said the US has a tech strategy for AI competition, but still lacks a clear people strategy for workers.
00:16:32.799 --> 00:16:43.759
So Ray's US is based by major players in tech and AI, including the Open AI Foundation, Anthropic, Amazon, and Microsoft, which are serving as anchor partners.
00:16:43.919 --> 00:16:55.919
So the nonprofit says it will build a national platform to advise state governors on how to prepare local workforces for AI-driven change with a focus on practical policy and training models.
00:16:56.240 --> 00:17:00.480
So far, uh there are some notable state partnerships already, right?
00:17:00.559 --> 00:17:03.200
Kind of hours after we got win of this new group.
00:17:03.360 --> 00:17:15.039
So the first state partnerships are all already underway with Arkansas, Connecticut, Maryland, and Utah split evenly between Republican and Democratic-led states.
00:17:15.200 --> 00:17:29.279
In Arkansas, Ray's US is working with governor uh Governor Sarah Huckabee Sanders on Arkansas launch, an AI-powered career navigation platform that would connect students and job seekers to personalized learning and employer-linked career paths.
00:17:29.440 --> 00:17:38.400
In Maryland, the group is working with Governor Wes Moore to expand its service year opportunities for recent high school graduates in fields like healthcare and education.
00:17:38.720 --> 00:17:51.359
Utah is a notable test because the state has been forced into the national spotlight over the build-out of AI data centers, showing how AI expansion can create both job and infrastructure tensions.
00:17:51.519 --> 00:18:05.359
So RAISE US says more states are expected to join in the coming months and also wants to test policies such as short time compensation and wage insurance that could soften the blow of job changes.
00:18:05.519 --> 00:18:11.599
So the effort comes as debate continues over how much AI will affect the labor market.
00:18:12.000 --> 00:18:31.599
As we know, we heard uh kind of some uh wild uh kind of predictions, such as that from Anthropic CEO Dario Mati, who originally said that AI would disrupt up to 50% of white-collar jobs, which obviously didn't happen, and he kind of had to adjust or update his stance on that.
00:18:31.680 --> 00:18:39.200
But regardless, we are already seeing in 2026 an unprecedented uh impact of AI on full-time jobs.
00:18:39.440 --> 00:18:58.240
So uh it's good to see a company come in and address the human side, which I've always said since I started this podcast three and a half years ago that AI implementation is much more about people and man, uh people management, change management than it is about you know traditional tech implementation or digital transformation.
00:18:58.400 --> 00:19:10.079
So, all right, our next piece of AI news we have a fresh accusation of Anthropic pointing their fingers at the Chinese saying you stole our models.
00:19:10.240 --> 00:19:25.759
So, according to C NBC, Anthropic sent a letter this month to the Senate banking committee accusing Alibaba of carrying out what it called the largest known distillation attack on Anthropic's AI models.
00:19:25.920 --> 00:19:42.240
So Anthropic said the activity involved about 28 million exchanges through roughly 25,000 fraudulent accounts between April and June, which it says points to a large-scale attempt to copy its model behavior.
00:19:42.400 --> 00:19:55.440
So the company told lawmakers that operators linked to Alibaba and its AI lab were behind the effort, framing the issue as a major example of AI model theft rather than normal competition.
00:19:55.519 --> 00:20:02.400
So distillation is a process, if you don't know, where a smaller model learns from the outputs of a stronger one.
00:20:02.559 --> 00:20:13.359
But Anthropic is accusing the Alibaba-linked effort of using that method to, well, just steal all of their work and infuse it into Alibaba's models.
00:20:13.599 --> 00:20:23.039
Anthropic said fighting this kind of AI copying will require coordination between government and industry, and that it wants Washington to help protect U.S.
00:20:23.200 --> 00:20:24.400
leadership.
00:20:24.720 --> 00:20:39.519
So, according to reports, the letter comes two months after the White House Office of Science and Technology Policy said it would help AI companies in the US detect and respond to industrial-scale distillation attempts.
00:20:39.680 --> 00:20:50.960
Anthropic said Alibaba's actions ignored the Trump administration's warnings, uh, making the disputes part of the bigger US-China technology and national security fight.
00:20:51.200 --> 00:21:00.720
So this is definitely a big story to keep a watch on because, well, this also impacts what models your company may or may not be using.
00:21:00.880 --> 00:21:06.720
So last week, make sure you go check out uh our start here series if you haven't already.
00:21:06.960 --> 00:21:09.599
But in episode uh which one was that?
00:21:09.759 --> 00:21:27.599
In episode 29 of the Start Here series, which was episode 804 of Everyday AI, we talked about a new open source model called GLM52, which may be the first time we have a Chinese open weight or open source model that is almost at the frontier level.
00:21:27.839 --> 00:21:29.839
So it's gonna be pretty interesting.
00:21:29.920 --> 00:21:41.119
And I think this uh probably deserves a larger conversation if the access to true frontier models is no longer actually democratized, right?
00:21:41.200 --> 00:21:56.720
If the 99.9% of US workers, at least right now, right, can't access Anthropics Fable, they can't access uh Anthropics mythos, they can't access right now, at least maybe that might change in a week or two.
00:21:56.880 --> 00:21:59.279
They can't access, you know, GPT 56 soul.
00:22:00.319 --> 00:22:04.880
So when you have these other Chinese models that are starting to get in that tier, right?
00:22:04.960 --> 00:22:10.160
I wouldn't say they're in the tier of GPT 5.6 and Mythos, but they're in the tier right below, right?
00:22:10.240 --> 00:22:14.240
Which is, well, models available to every single person, right?
00:22:14.400 --> 00:22:18.720
So that's GPT-5.5 and Opus 4.8 from Anthropic.
00:22:18.960 --> 00:22:20.960
So as you get these Chinese models, right?
00:22:21.119 --> 00:22:31.039
Whether they're from Alibaba, uh, they're from ZAI, whoever they are, that are starting to get in that tier of the best publicly available models, it is an important conversation to keep an eye on.
00:22:31.119 --> 00:22:50.000
So distillation, it's it's the combination of distillation and maybe these frontier systems only being available to the select few uh creates for a very interesting and mixy uh uh sorry, a potentially um explosive combination, right?
00:22:50.160 --> 00:22:58.559
In terms of what models your company should use and well, what it means for revenue of companies like anthropic who are preparing to go public.
00:22:59.279 --> 00:22:59.759
All right.
00:23:00.079 --> 00:23:04.319
Our last big piece of AI news this week is chip news.
00:23:04.480 --> 00:23:18.079
Yeah, because open AI and Broadcom have unveiled Jalapeno, a custom chip meant to run large language models more efficiently, with deployment expected to happen later this year.
00:23:18.240 --> 00:23:44.319
So the big picture goal for OpenAI and Broadcom was to make an AI systems uh to make their AI systems cheaper and faster and less power hungry to run at scale, which matters a lot because energy and computing costs are becoming major limits for AI companies and their ability to provide uh frontier AI models to the broader enterprise um lens.
00:23:44.880 --> 00:23:59.680
So, with this new chip partnership, OpenAI's first, open AI says jalapeno is built specifically for AI inference, meaning it is designed to help models answer questions and serve users rather than train them, right?
00:23:59.839 --> 00:24:06.079
So, obviously, on the training side, Nvidia's GPUs have been the industry standard for a long time.
00:24:06.400 --> 00:24:17.359
This chip is specifically built for inference, like all of the back and forth that open AI needs when uh up to a billion users each week are using their services.
00:24:17.599 --> 00:24:27.119
So the companies say the chip is aimed at the hardest parts of the AI service, including speed, memory use, networking, and reducing wasted compute.
00:24:27.279 --> 00:24:37.279
So OpenAI and Broadcom claim the chip will be more efficient than current leading hardware, but they have not shared benchmarks or hard performance numbers yet.
00:24:37.519 --> 00:24:46.079
Images shared by the company suggest a very large chip with multiple memory modules, which points to a serious effort to handle heavy AI workloads.
00:24:46.240 --> 00:24:52.400
So the companies say early samples are already working in the lab, which is a crazy fast turnaround.
00:24:52.480 --> 00:24:55.680
And OpenAI says it is testing real workloads on the chips now.
00:24:55.920 --> 00:25:10.000
OpenAI and Broadcom also say the chip was developed in just nine months, which is much faster than typical uh typical chip development and shows how quickly AI infrastructure is being pushed forward.
00:25:10.240 --> 00:25:15.519
All right, so that's a wrap for our major big stories of the week.
00:25:16.319 --> 00:25:20.319
But let's quickly move into what's new and what's next.
00:25:20.480 --> 00:25:28.559
So this is a bullet point roundup of everything else noteworthy that either happened last week or is happening this week.
00:25:28.720 --> 00:25:34.160
So it's a mixture of smaller news stories, maybe some leaks and rumors, but let's get straight into it.
00:25:34.319 --> 00:25:35.680
Fast bullet point style.
00:25:35.759 --> 00:25:36.160
Here we go.
00:25:36.319 --> 00:25:41.920
So nearly 400 newspapers have sued OpenAI in Microsoft over content.
00:25:42.160 --> 00:25:48.240
Qualcomm and Meta announced a multi-generation data center uh CPU partnership.
00:25:48.400 --> 00:25:54.000
Uh Chevron signed a 20-year Microsoft power deal for AI data centers.
00:25:54.079 --> 00:26:05.920
Uh, Meta launched a$299 pair of AI glasses uh in line with SLR Luxotica, the famous ibrand maker that I probably did not pronounce correctly.
00:26:06.000 --> 00:26:11.839
Uh Google reportedly limited Meta's access to Gemini AI models over capacity shortages.
00:26:12.079 --> 00:26:21.200
Elon Musk's says that XAI's new 1.5 trillion parameter Groc 4.5 has begun private beta testing.
00:26:21.279 --> 00:26:28.720
So we'll see if they have to go through the normal uh government uh hoops uh jumping through those or if they'll be able to release it.
00:26:28.960 --> 00:26:34.079
Uh Micron and Anthropic sign a memory deal to help scale Claude.
00:26:34.400 --> 00:26:42.160
Apple's VP4 Vision Pro and smart glasses is leaving for OpenAI's hardware unit, according to reports.
00:26:42.319 --> 00:26:50.319
Uh leaks show that OpenAI is testing computer use with Excel and PowerPoint add-ins in codecs, which will be an interesting update.
00:26:50.480 --> 00:26:57.039
Uh, Google, inside of Notebook LM, added fully customizable flashcards for users to edit and share.
00:26:57.200 --> 00:27:07.440
OpenAI reportedly is weighing a 2027 IPO instead of 2026 after some recent SpaceX volatility.
00:27:07.599 --> 00:27:16.480
In an internal study, OpenAI said Codecs accounted for 99.8% of internal weekly AI output tokens, which is a wild percentage.
00:27:16.640 --> 00:27:22.720
Almost 100% of their output tokens internally are in codecs, not in their chat GPT service.
00:27:22.960 --> 00:27:29.039
OpenAI released daybreak cyber tools, including GPT-5.5 cyber and codec security updates.
00:27:29.200 --> 00:27:35.920
OpenAI released a refreshed GPT 5.5 instant model with better conversation and handling and constraints.
00:27:36.160 --> 00:27:42.559
Google launched Google Finance in general availability with portfolio tracking and AI research tools.
00:27:42.720 --> 00:27:47.519
Google released Gemini for home smart speakers for natural voice conversations.
00:27:47.759 --> 00:27:55.759
Anthropic launched Claude Tag in Slack, which is kind of like a manager per Slack channel that can do anything that uh Claude can do.
00:27:56.000 --> 00:28:01.119
Canva released Canva Grow 2.0 for AI ad creation, publishing, and optimization.
00:28:01.359 --> 00:28:12.240
Google added Select from Screen in Chrome's Gemini sidebar after they updated their Gemini 3.5 uh flash to have computer use capabilities.
00:28:12.400 --> 00:28:24.079
Microsoft added Copilot skills in Excel for reputable task automation and a new digital ocean plugin for codecs, entered public preview for remote cloud dev machines.
00:28:24.240 --> 00:28:25.519
All right, that's a lot.
00:28:25.599 --> 00:28:31.279
And I can almost guarantee you this week is going to move even faster than last week.
00:28:31.359 --> 00:28:36.000
Because y'all, this everything with OpenAI and Anthropic in terms of GBD5.
00:28:39.039 --> 00:28:47.759
The government uh kind of asking them to hold it back, the re-release, the slow re-release of Anthropic Mythos, all of this just happened Friday.
00:28:47.839 --> 00:28:52.000
So I expect a ton of new developments that you need to be paying attention to.
00:28:52.319 --> 00:29:00.400
So don't spend hours every single day sitting around your boardroom talking about AI strategy and looking at that day's news, right?
00:29:00.559 --> 00:29:02.000
Just let me do that for you.
00:29:02.160 --> 00:29:05.440
Then join us on Mondays as we bring you the AI news that matters.
00:29:05.599 --> 00:29:08.000
If you're new here, on Wednesdays, we go hands-on.
00:29:08.160 --> 00:29:09.359
We usually do a demo.
00:29:09.440 --> 00:29:14.799
Uh, so make sure you check out today's newsletter so you can let us know what you want to see on that show.
00:29:14.960 --> 00:29:19.759
On Fridays, we go over fresh uh AI features that are available to everyone.
00:29:19.920 --> 00:29:23.119
And then on Tuesdays and Thursdays, we rotate the type of show.
00:29:23.279 --> 00:29:28.000
So uh please, if you haven't already, subscribe to the show on Spotify and Apple Podcasts.
00:29:28.079 --> 00:29:32.319
Then go to your everydayai.com, sign up for the free daily newsletter.
00:29:32.400 --> 00:29:33.119
Thanks for tuning in.
00:29:33.200 --> 00:29:36.160
We hope to see you back tomorrow and every day for more everyday AI.
00:29:36.240 --> 00:29:37.279
Thanks, y'all.
