1
00:00:00,000 -->
00:00:02,280
I worry about the general person
2
00:00:02,280 -->
00:00:06,560
that doesn't quite understand what they're getting into.
3
00:00:06,560 -->
00:00:11,560
And we all know Meta does not have the best track record
4
00:00:11,560 -->
00:00:14,560
of user privacy, of security.
5
00:00:14,560 -->
00:00:15,920
These are just the facts.
6
00:00:15,920 -->
00:00:18,640
You could go back on the archive of this podcast.
7
00:00:18,640 -->
00:00:22,000
Like going back to Cambridge Analytica
8
00:00:22,000 -->
00:00:23,880
and even before that,
9
00:00:23,880 -->
00:00:27,220
that Meta is a very difficult company to trust
10
00:00:27,220 -->
00:00:29,120
for many different reasons.
11
00:00:29,120 -->
00:00:31,080
This week on Shared Security,
12
00:00:31,080 -->
00:00:34,620
your AI assistant wants to clear your inbox,
13
00:00:34,620 -->
00:00:36,840
shuffle your calendar, shop the web,
14
00:00:36,840 -->
00:00:41,380
and generally become the world's most eager digital intern.
15
00:00:41,380 -->
00:00:44,520
That's really cute until it's holding the keys
16
00:00:44,520 -->
00:00:46,860
to your entire online life.
17
00:00:46,860 -->
00:00:49,480
Well, Meta's new Muse agent
18
00:00:49,480 -->
00:00:52,100
is built to work across connected apps,
19
00:00:52,100 -->
00:00:55,240
but a reported zero day gives that promise
20
00:00:55,240 -->
00:00:58,340
a very real stress test.
21
00:00:58,380 -->
00:01:01,340
So this week we're asking where helpful ends
22
00:01:01,340 -->
00:01:04,380
and a giant attack surface begins.
23
00:01:04,380 -->
00:01:06,960
So we'll look at Meta's cloud virtual machine,
24
00:01:06,960 -->
00:01:10,140
credential isolation and approval claims,
25
00:01:10,140 -->
00:01:11,540
and what can go wrong
26
00:01:11,540 -->
00:01:14,900
when a compromised agent has delegated authority.
27
00:01:14,900 -->
00:01:19,480
So Scott, if an AI gets a master key to your digital life,
28
00:01:19,480 -->
00:01:23,460
what guardrails would make you trust it with anything?
29
00:01:23,460 -->
00:01:25,300
That's a good question, Tom.
30
00:01:25,300 -->
00:01:26,220
I'm not sure.
31
00:01:26,220 -->
00:01:28,500
Well, certainly I'm a bad person to ask
32
00:01:28,500 -->
00:01:31,460
because I've not yet employed any agents or assistants
33
00:01:31,460 -->
00:01:34,300
and I'm waiting to see how the dust settles.
34
00:01:34,300 -->
00:01:37,500
So I'm not really a good person to ask that too,
35
00:01:37,500 -->
00:01:41,140
unless you are a little on the conservative side
36
00:01:41,140 -->
00:01:42,220
when it comes to technology.
37
00:01:42,220 -->
00:01:45,740
So as you know, I mean, I'm a technology fan
38
00:01:45,740 -->
00:01:48,380
but I've always had my reservations
39
00:01:48,380 -->
00:01:50,660
about where AI is going.
40
00:01:50,660 -->
00:01:52,500
I think it brings me back.
41
00:01:52,500 -->
00:01:55,740
I keep seeing these little video shorts on YouTube
42
00:01:56,260 -->
00:01:57,380
of people talking about it.
43
00:01:57,380 -->
00:02:00,140
And it's one I saw today was really kind of reminded me
44
00:02:00,140 -->
00:02:03,900
of something we've sort of lost track of.
45
00:02:03,900 -->
00:02:06,580
And that is even the companies
46
00:02:06,580 -->
00:02:08,060
that are creating these things
47
00:02:08,060 -->
00:02:10,140
don't know exactly how they work.
48
00:02:10,140 -->
00:02:11,300
No, they don't.
49
00:02:11,300 -->
00:02:13,500
It's kind of shocking.
50
00:02:13,500 -->
00:02:16,020
And I don't know if you saw the news the other day
51
00:02:16,020 -->
00:02:20,540
about all of the AI tech CEOs met with Donald Trump
52
00:02:20,540 -->
00:02:24,780
at the White House to create this like agreement
53
00:02:24,780 -->
00:02:29,780
that their companies will be regulating this on their own
54
00:02:30,220 -->
00:02:31,420
which is kind of hilarious
55
00:02:31,420 -->
00:02:33,620
because the government doesn't want to regulate it.
56
00:02:33,620 -->
00:02:37,180
And I saw the clip of Dario Amade,
57
00:02:37,180 -->
00:02:38,780
he was very, very awkward.
58
00:02:38,780 -->
00:02:41,420
Like, I mean, he's just a socially awkward individual.
59
00:02:41,420 -->
00:02:43,580
I mean, which tech CEO isn't, right?
60
00:02:43,580 -->
00:02:45,300
But he was kind of saying like,
61
00:02:45,300 -->
00:02:48,780
well, we're still talking about it about the dangers guys
62
00:02:48,780 -->
00:02:50,700
and we haven't agreed yet.
63
00:02:50,700 -->
00:02:52,460
And then, you know, Mark Zuckerberg
64
00:02:52,460 -->
00:02:54,260
and Trump are kind of snickering, you know?
65
00:02:54,260 -->
00:02:59,620
And it's like, they have no idea like how serious this is, right?
66
00:02:59,620 -->
00:03:00,740
No, no, no.
67
00:03:00,740 -->
00:03:02,620
So I'm not about to trust any of these things.
68
00:03:02,620 -->
00:03:03,780
Yeah.
69
00:03:03,780 -->
00:03:06,300
It's, you know, we're delegating our authority
70
00:03:06,300 -->
00:03:07,700
for sending emails.
71
00:03:07,700 -->
00:03:09,980
I mean, in the topic you're talking about today,
72
00:03:09,980 -->
00:03:11,820
I think it's less powerful
73
00:03:11,820 -->
00:03:15,100
because it's just an assistant that I guess is teeing stuff up
74
00:03:15,100 -->
00:03:17,300
and asking you if you want to send it
75
00:03:17,300 -->
00:03:18,580
or waiting for you to do it.
76
00:03:18,580 -->
00:03:22,020
But, you know, the whole industry seems to be moving
77
00:03:22,060 -->
00:03:25,340
towards more automation, which just in a funny way
78
00:03:25,340 -->
00:03:27,500
reminds me of the, there was a movie called Wally,
79
00:03:27,500 -->
00:03:28,340
do you remember that?
80
00:03:28,340 -->
00:03:29,180
Yeah, of course.
81
00:03:29,180 -->
00:03:31,660
Where the guys end up sitting on chairs getting moved.
82
00:03:31,660 -->
00:03:32,500
Yes.
83
00:03:32,500 -->
00:03:33,980
Nobody's spending any effort
84
00:03:33,980 -->
00:03:35,580
either physically or mentally anymore.
85
00:03:35,580 -->
00:03:39,380
That's our target objective seems to be to do nothing.
86
00:03:39,380 -->
00:03:40,220
Yeah.
87
00:03:40,220 -->
00:03:41,460
We're living Wally in real life.
88
00:03:41,460 -->
00:03:42,300
We are.
89
00:03:42,300 -->
00:03:43,380
That's exactly what is happening.
90
00:03:43,380 -->
00:03:44,700
So talking about Muse,
91
00:03:44,700 -->
00:03:48,460
so this is Metta's really first attempt
92
00:03:48,500 -->
00:03:53,500
at introducing AI agents to a massive population of people.
93
00:03:54,180 -->
00:03:58,180
And AI agents, this isn't something that's brand new, right?
94
00:03:58,180 -->
00:04:00,980
We've been talking about agents for quite a while now.
95
00:04:00,980 -->
00:04:04,020
In fact, I'm using an AI agent,
96
00:04:04,020 -->
00:04:06,420
automate some of the podcast workflow
97
00:04:06,420 -->
00:04:08,500
and maybe we'll do an episode about that one day.
98
00:04:08,500 -->
00:04:10,780
But I've actually found it very useful,
99
00:04:10,780 -->
00:04:14,720
but I would say that it's something that I had to train
100
00:04:14,720 -->
00:04:18,340
after many weeks and months of fine tuning
101
00:04:18,340 -->
00:04:23,060
and of course, human loop oversight of what it's doing
102
00:04:23,060 -->
00:04:26,500
that I'm actually comfortable letting it do the things
103
00:04:26,500 -->
00:04:28,080
that it does for me.
104
00:04:28,080 -->
00:04:31,260
And even when it does those things for me,
105
00:04:31,260 -->
00:04:33,620
it doesn't always do it right.
106
00:04:33,620 -->
00:04:37,180
And even though I have written very specific,
107
00:04:37,180 -->
00:04:41,980
deterministic playbooks for things around the podcast
108
00:04:41,980 -->
00:04:43,580
in terms of like my editing
109
00:04:43,580 -->
00:04:46,700
and some of that manual work,
110
00:04:46,700 -->
00:04:49,700
I find that it still gets that wrong at times.
111
00:04:49,700 -->
00:04:51,980
Maybe 90% of the time it's right,
112
00:04:51,980 -->
00:04:54,700
but you still have to babysit it and oversee it.
113
00:04:54,700 -->
00:04:59,200
And I think that's the thing with this Muse agent.
114
00:04:59,200 -->
00:05:00,860
Of course, Facebook and Metta,
115
00:05:00,860 -->
00:05:03,220
they're saying that, oh yeah, we've tested this
116
00:05:03,220 -->
00:05:06,300
and we verified it and we have all these guardrails
117
00:05:06,300 -->
00:05:07,580
and these controls around it.
118
00:05:07,580 -->
00:05:09,260
And we can talk about some of those things.
119
00:05:09,260 -->
00:05:12,420
Like one, each agent runs in its own
120
00:05:12,420 -->
00:05:15,500
virtual segmented environment, right?
121
00:05:15,500 -->
00:05:16,660
So other agents are not.
122
00:05:16,660 -->
00:05:18,580
Probably ironclad sandbox, right?
123
00:05:18,580 -->
00:05:19,420
Yes, yes.
124
00:05:19,420 -->
00:05:20,540
They're thinking in and out of the chint.
125
00:05:20,540 -->
00:05:23,180
And probably using military grade encryption.
126
00:05:23,180 -->
00:05:24,300
And of course, right?
127
00:05:24,300 -->
00:05:26,060
Yeah, all the things in the troops
128
00:05:26,060 -->
00:05:28,380
that we've talked about that can be easily hacked, right?
129
00:05:28,380 -->
00:05:30,240
But I'm giving them the benefit of the doubt here.
130
00:05:30,240 -->
00:05:33,260
I mean, no doubt that they have tested this.
131
00:05:33,260 -->
00:05:34,420
But what I find interesting,
132
00:05:34,420 -->
00:05:36,180
and we'll link in the show notes the article
133
00:05:36,180 -->
00:05:37,940
from Metta itself.
134
00:05:37,940 -->
00:05:42,020
And they literally say that we know
135
00:05:42,020 -->
00:05:44,940
that Muse will not be right.
136
00:05:44,940 -->
00:05:47,100
Like we know it's going to make mistakes.
137
00:05:47,100 -->
00:05:50,500
We know there is going to be security vulnerabilities.
138
00:05:50,500 -->
00:05:52,380
We know it's gonna attack other websites.
139
00:05:52,380 -->
00:05:54,620
But, and I find it's hilarious.
140
00:05:54,620 -->
00:05:56,860
And they're like, well, now it's publicly available
141
00:05:56,860 -->
00:06:00,460
for anybody with a Metta account to use.
142
00:06:00,460 -->
00:06:04,220
So while on one hand you say, yeah, it's safe.
143
00:06:04,220 -->
00:06:05,300
We've tested it.
144
00:06:05,300 -->
00:06:06,620
And then on the other,
145
00:06:06,620 -->
00:06:09,820
it may do things that you didn't want it to do.
146
00:06:09,820 -->
00:06:11,260
And there could be security issues.
147
00:06:11,260 -->
00:06:13,900
Like we just saw, there was a zero day
148
00:06:13,940 -->
00:06:16,100
that was found recently in,
149
00:06:16,100 -->
00:06:17,780
I believe it was around the workflow
150
00:06:17,780 -->
00:06:20,440
of how this agent works.
151
00:06:20,440 -->
00:06:23,780
And so these things are gonna naturally happen.
152
00:06:23,780 -->
00:06:25,820
But despite all the controls,
153
00:06:25,820 -->
00:06:28,700
and we're not gonna go into every detail
154
00:06:28,700 -->
00:06:29,900
around those guardrails,
155
00:06:29,900 -->
00:06:32,940
but I think they've taken the safest approach
156
00:06:32,940 -->
00:06:35,820
that they can to let people use this.
157
00:06:35,820 -->
00:06:38,900
Now, there's all kinds of controversy already about this.
158
00:06:38,900 -->
00:06:41,580
I read a 404 media article recently
159
00:06:41,620 -->
00:06:45,820
about how Muse is actually using human agents.
160
00:06:45,820 -->
00:06:47,980
So when you tell your Muse agent
161
00:06:47,980 -->
00:06:51,060
to like schedule my doctor appointment for me,
162
00:06:51,060 -->
00:06:53,500
the agent doesn't go to like to your doctor's website
163
00:06:53,500 -->
00:06:55,940
and actually like go through the pages
164
00:06:55,940 -->
00:06:57,580
and schedule the appointment.
165
00:06:57,580 -->
00:06:59,460
It goes through the halfway around the world.
166
00:06:59,460 -->
00:07:00,900
Yeah, yeah, exactly.
167
00:07:00,900 -->
00:07:03,580
It actually gets a human in the loop
168
00:07:03,580 -->
00:07:05,740
to actually make the phone call.
169
00:07:05,740 -->
00:07:08,740
But of course, Metta didn't disclose this
170
00:07:08,740 -->
00:07:09,580
so lower or something.
171
00:07:09,580 -->
00:07:11,420
That's so funny really.
172
00:07:11,660 -->
00:07:14,220
It's come full circle to what people used to joke about.
173
00:07:14,220 -->
00:07:18,340
Like AI is really just a bunch of people in a sweatshop.
174
00:07:18,340 -->
00:07:19,180
Yeah, it is.
175
00:07:19,180 -->
00:07:21,700
There's always people behind the scenes doing things.
176
00:07:21,700 -->
00:07:23,340
And I've actually seen this
177
00:07:23,340 -->
00:07:26,700
with a lot of these AI startups too
178
00:07:26,700 -->
00:07:29,500
is they all claim the big one right now
179
00:07:29,500 -->
00:07:32,260
is in our industry is AI pen testing,
180
00:07:32,260 -->
00:07:34,820
fully autonomous pen testing.
181
00:07:34,820 -->
00:07:38,860
And it looks and appears to be an automated agent
182
00:07:38,860 -->
00:07:40,380
but behind the scene,
183
00:07:40,380 -->
00:07:43,460
there's actually a pen tester driving it.
184
00:07:43,460 -->
00:07:45,060
Well, that's the lean startup model, right?
185
00:07:45,060 -->
00:07:46,620
Is it do stuff that's not scalable
186
00:07:46,620 -->
00:07:48,860
until you find a market, right?
187
00:07:48,860 -->
00:07:49,980
I think it's interesting to me
188
00:07:49,980 -->
00:07:52,020
that since this is the first,
189
00:07:52,020 -->
00:07:56,700
what I would say mainstream AI agent
190
00:07:56,700 -->
00:08:00,020
that is available to the masses
191
00:08:00,020 -->
00:08:02,140
because let's be honest, I mean Metta
192
00:08:02,140 -->
00:08:06,340
because they own WhatsApp, Instagram and Facebook.
193
00:08:06,380 -->
00:08:09,580
Like you're talking millions and millions of people
194
00:08:09,580 -->
00:08:11,620
that could be using this now.
195
00:08:11,620 -->
00:08:15,580
And where AI agents in the past and just the recent past,
196
00:08:15,580 -->
00:08:18,300
like literally a couple of months ago
197
00:08:18,300 -->
00:08:21,260
were either corporations and businesses using it
198
00:08:21,260 -->
00:08:24,340
through like Claude Cowork or co-pilot
199
00:08:24,340 -->
00:08:26,980
or open AI has it's now,
200
00:08:26,980 -->
00:08:28,900
I think they call it work or something
201
00:08:28,900 -->
00:08:31,060
where these agents are going out
202
00:08:31,060 -->
00:08:32,820
and doing things on your behalf.
203
00:08:32,820 -->
00:08:34,420
There's a whole open source community
204
00:08:34,420 -->
00:08:35,700
like I use Hermes,
205
00:08:35,700 -->
00:08:39,380
which is a very popular agent harness that's out there.
206
00:08:39,380 -->
00:08:41,020
There's many others as well.
207
00:08:41,020 -->
00:08:45,020
We've talked about Open Claw, all these things.
208
00:08:45,020 -->
00:08:47,900
So, but they've been kind of segmented
209
00:08:47,900 -->
00:08:50,740
to the tech community, right?
210
00:08:50,740 -->
00:08:54,100
But not the general user like my mom
211
00:08:54,100 -->
00:08:58,340
or your grandma can now use a muse agent
212
00:08:58,340 -->
00:09:00,420
and what does that mean?
213
00:09:00,420 -->
00:09:03,300
I think it's wild.
214
00:09:03,340 -->
00:09:06,220
Mobile apps are just part of everyday life now.
215
00:09:06,220 -->
00:09:09,420
Banking, healthcare, shopping, entertainment, you name it.
216
00:09:09,420 -->
00:09:11,940
And with that comes a lot of trust
217
00:09:11,940 -->
00:09:14,100
because users are putting their personal data
218
00:09:14,100 -->
00:09:16,180
directly into your app.
219
00:09:16,180 -->
00:09:17,300
But here's the reality.
220
00:09:17,300 -->
00:09:19,500
Mobile apps are a growing target.
221
00:09:19,500 -->
00:09:22,660
A recent survey found that 72% of organizations
222
00:09:22,660 -->
00:09:26,260
experienced a mobile app security incident last year
223
00:09:26,260 -->
00:09:29,980
and 92% say threats are only increasing.
224
00:09:29,980 -->
00:09:31,980
And the way attackers are going after apps
225
00:09:32,020 -->
00:09:33,580
is pretty sophisticated.
226
00:09:33,580 -->
00:09:35,260
They're reverse engineering them,
227
00:09:35,260 -->
00:09:38,300
modifying them and redistributing fake versions
228
00:09:38,300 -->
00:09:39,860
through phishing campaigns,
229
00:09:39,860 -->
00:09:43,220
side loading and even third party app stores.
230
00:09:43,220 -->
00:09:45,140
So from a user's perspective,
231
00:09:45,140 -->
00:09:47,860
everything can look completely legitimate.
232
00:09:47,860 -->
00:09:49,580
That's why taking a proactive approach
233
00:09:49,580 -->
00:09:52,060
to mobile app security really matters.
234
00:09:52,060 -->
00:09:53,580
You wanna stay ahead of these threats,
235
00:09:53,580 -->
00:09:56,300
not react after the damage is done.
236
00:09:56,300 -->
00:09:58,260
This is where Guard Square comes in.
237
00:09:58,260 -->
00:09:59,700
They provide advanced protection
238
00:09:59,740 -->
00:10:02,100
for both Android and iOS apps,
239
00:10:02,100 -->
00:10:04,340
along with automated security testing
240
00:10:04,340 -->
00:10:06,620
to catch vulnerabilities early
241
00:10:06,620 -->
00:10:08,540
and real-time threat monitoring
242
00:10:08,540 -->
00:10:11,220
so you can actually see what's happening out there.
243
00:10:11,220 -->
00:10:13,660
If your mobile app is critical to your business
244
00:10:13,660 -->
00:10:15,220
and it probably is,
245
00:10:15,220 -->
00:10:17,620
this is something worth paying attention to.
246
00:10:17,620 -->
00:10:20,420
You can learn more at guardsquare.com.
247
00:10:20,420 -->
00:10:24,830
That's guardsquare.com.
248
00:10:24,830 -->
00:10:26,230
It's wild for sure.
249
00:10:26,230 -->
00:10:29,670
I mean, another thing that comes to mind is liability,
250
00:10:29,870 -->
00:10:33,070
I mean, as we know, as you said just now,
251
00:10:33,070 -->
00:10:34,590
these things are gonna make mistakes
252
00:10:34,590 -->
00:10:36,750
and how costly are those mistakes gonna be
253
00:10:36,750 -->
00:10:39,270
or how dangerous or physically
254
00:10:39,270 -->
00:10:40,830
are those things gonna be, right?
255
00:10:40,830 -->
00:10:44,390
So it reminds me again of going back
256
00:10:44,390 -->
00:10:47,830
to the early market discussion
257
00:10:47,830 -->
00:10:49,070
around self-driving cars.
258
00:10:49,070 -->
00:10:50,270
Well, who's gonna be liable
259
00:10:50,270 -->
00:10:52,870
when the car kills somebody?
260
00:10:52,870 -->
00:10:54,430
At least with self-driving cars,
261
00:10:54,430 -->
00:10:55,590
you've got a physical object.
262
00:10:55,590 -->
00:10:57,630
You've got, you had software
263
00:10:57,630 -->
00:10:58,910
that was physically running in it
264
00:10:59,030 -->
00:11:02,990
and you could kind of identify who the owner was
265
00:11:02,990 -->
00:11:04,510
or who agreed to this stuff.
266
00:11:04,510 -->
00:11:07,190
But now everything is so ethereal, right?
267
00:11:07,190 -->
00:11:08,830
You don't really know,
268
00:11:08,830 -->
00:11:10,230
first of all, you don't know where it's running.
269
00:11:10,230 -->
00:11:13,670
You don't know whose authority it's running on
270
00:11:13,670 -->
00:11:16,190
and it's gonna be very complicated, I think,
271
00:11:16,190 -->
00:11:17,910
for this to get ironed out
272
00:11:17,910 -->
00:11:20,350
and if it ever gets ironed out in courts.
273
00:11:20,350 -->
00:11:24,310
It's sort of like when we accept shrink wrap software
274
00:11:24,310 -->
00:11:27,270
not really knowing if it's safe or not
275
00:11:27,270 -->
00:11:28,110
or who's liable
276
00:11:28,110 -->
00:11:30,910
but in the end the software company's got
277
00:11:30,910 -->
00:11:34,350
pretty much protected against anything
278
00:11:34,350 -->
00:11:36,230
other than the cost of the software itself.
279
00:11:36,230 -->
00:11:39,590
So this is gonna be a huge thing, I think,
280
00:11:39,590 -->
00:11:41,230
in terms of liability.
281
00:11:41,230 -->
00:11:43,510
Yeah, I guess I worry mostly
282
00:11:43,510 -->
00:11:46,390
about just the general user, right?
283
00:11:46,390 -->
00:11:48,830
That is not technically savvy.
284
00:11:48,830 -->
00:11:52,190
That doesn't really understand the power
285
00:11:52,190 -->
00:11:55,870
that the Muse agent actually has.
286
00:11:55,910 -->
00:11:59,950
I mean, this thing can answer your emails for you.
287
00:11:59,950 -->
00:12:01,870
It can schedule events.
288
00:12:01,870 -->
00:12:04,030
It can pay your bills.
289
00:12:04,030 -->
00:12:06,710
It can do very, very powerful things
290
00:12:06,710 -->
00:12:09,270
that people don't quite realize
291
00:12:09,270 -->
00:12:11,950
if something goes wrong and it probably will.
292
00:12:11,950 -->
00:12:14,030
Yeah, that wasn't me, that was the agent.
293
00:12:14,030 -->
00:12:14,870
Yeah, exactly.
294
00:12:14,870 -->
00:12:16,510
I've seen this with my own agent
295
00:12:16,510 -->
00:12:19,150
just on a simple thing like a podcast workflow.
296
00:12:19,150 -->
00:12:21,550
Make mistakes, I'll be honest.
297
00:12:21,550 -->
00:12:23,990
I will not trust that with my finances
298
00:12:23,990 -->
00:12:27,150
or making decisions and things
299
00:12:27,150 -->
00:12:31,070
that I as a human need to make
300
00:12:31,070 -->
00:12:33,590
but I worry about the general person
301
00:12:33,590 -->
00:12:37,870
that doesn't quite understand what they're getting into
302
00:12:37,870 -->
00:12:42,870
and we all know Meta does not have the best track record
303
00:12:42,870 -->
00:12:45,870
of user privacy, of security.
304
00:12:45,870 -->
00:12:47,190
These are just the facts.
305
00:12:47,190 -->
00:12:49,150
Like we've been, you could go back
306
00:12:49,150 -->
00:12:51,150
on the archive of this podcast.
307
00:12:51,150 -->
00:12:51,990
Yeah, yep, yep.
308
00:12:52,030 -->
00:12:54,470
Like going back to Cambridge Analytica
309
00:12:54,470 -->
00:12:56,310
and even before that,
310
00:12:56,310 -->
00:12:59,670
that Meta is a very difficult company to trust
311
00:12:59,670 -->
00:13:01,270
for many different reasons.
312
00:13:01,270 -->
00:13:03,310
Well, absolutely, it doesn't take too much, right?
313
00:13:03,310 -->
00:13:07,070
To imagine the discussion during early product roadmaps
314
00:13:07,070 -->
00:13:09,830
to say, you know, how much privacy are we gonna,
315
00:13:09,830 -->
00:13:11,910
you know, how much privacy protection
316
00:13:11,910 -->
00:13:12,870
are we gonna put into this?
317
00:13:12,870 -->
00:13:14,350
Well, our business model,
318
00:13:14,350 -->
00:13:16,310
we may wanna access that data someday, right?
319
00:13:16,310 -->
00:13:20,630
So it's clearly gonna be not as tight as we want.
320
00:13:20,630 -->
00:13:23,030
I mean, there's millions and millions of dollars at stake
321
00:13:23,030 -->
00:13:27,550
and this is obviously a way for Meta to make even more money
322
00:13:27,550 -->
00:13:30,830
and tie this to some subscription or, you know,
323
00:13:30,830 -->
00:13:31,990
something eventually.
324
00:13:31,990 -->
00:13:34,270
We've always said this about social media,
325
00:13:34,270 -->
00:13:36,950
about anything that these social media companies do.
326
00:13:36,950 -->
00:13:38,230
The end of the day,
327
00:13:38,230 -->
00:13:40,270
they are trying to find ways to make money
328
00:13:40,270 -->
00:13:41,870
off of your data.
329
00:13:41,870 -->
00:13:42,710
Yep.
330
00:13:42,710 -->
00:13:43,630
Plain and simple.
331
00:13:43,630 -->
00:13:44,670
That is their business model.
332
00:13:44,670 -->
00:13:45,870
It is, it is.
333
00:13:45,870 -->
00:13:48,350
So I'm interested to hear from all of you.
334
00:13:48,390 -->
00:13:51,030
Are you somebody that's going to try out Muse?
335
00:13:51,030 -->
00:13:54,430
Do you have friends or family members already using Muse?
336
00:13:54,430 -->
00:13:55,270
Let us know.
337
00:13:55,270 -->
00:13:58,550
We'd love to hear your thoughts and your feedback
338
00:13:58,550 -->
00:14:01,430
if this is something that scares the hell out of you
339
00:14:01,430 -->
00:14:05,030
or you are all in and you're using it every day.
340
00:14:05,030 -->
00:14:06,150
We'd love to know.
341
00:14:06,150 -->
00:14:07,030
Yes.
342
00:14:07,030 -->
00:14:09,030
So Scott, it sounds like you have an update
343
00:14:09,030 -->
00:14:11,590
about your digital legacy project.
344
00:14:11,590 -->
00:14:13,950
Yes, the digital legacy tree book.
345
00:14:14,830 -->
00:14:15,990
We're progressing.
346
00:14:15,990 -->
00:14:17,790
I'm probably a few weeks away hopefully
347
00:14:17,790 -->
00:14:19,190
from a Kindle version of it.
348
00:14:19,190 -->
00:14:23,870
So anybody who wants to read an early revision of it
349
00:14:23,870 -->
00:14:25,910
before I publish it can find it.
350
00:14:25,910 -->
00:14:29,030
I have the domain now digitallegacytree.com.
351
00:14:29,030 -->
00:14:31,310
So that's easy to find and you can just go there
352
00:14:31,310 -->
00:14:34,950
and say, start reading and send me your feedback in line.
353
00:14:34,950 -->
00:14:35,790
It's fun.
354
00:14:35,790 -->
00:14:36,630
It's easy to do.
355
00:14:36,630 -->
00:14:39,670
So happy to hear from people on that.
356
00:14:39,670 -->
00:14:43,070
But I mean, just today's discussion makes me think,
357
00:14:43,070 -->
00:14:47,030
first of all, the book itself is all about helping people
358
00:14:47,030 -->
00:14:49,150
ensure that their loved ones can access
359
00:14:49,150 -->
00:14:50,990
the important parts of their digital life
360
00:14:50,990 -->
00:14:52,030
when it matters the most.
361
00:14:52,030 -->
00:14:54,310
So when you die or when you're incapacitated
362
00:14:54,310 -->
00:14:57,910
or any other situations where you might be sort of
363
00:14:57,910 -->
00:15:00,330
long-term not able to access your own stuff.
364
00:15:00,330 -->
00:15:04,630
And certainly the idea of having agents
365
00:15:04,630 -->
00:15:07,070
with your identity acting on behalf of you
366
00:15:07,070 -->
00:15:10,110
which presumably will continue on after you die
367
00:15:10,110 -->
00:15:12,350
if you don't turn it off proactively.
368
00:15:12,350 -->
00:15:14,510
If you don't proactively say, I'm gonna die,
369
00:15:14,510 -->
00:15:15,910
stop doing this stuff, right?
370
00:15:16,910 -->
00:15:18,710
So that's an important discussion
371
00:15:18,710 -->
00:15:20,870
and I'd like to hear from people on whether or not
372
00:15:20,870 -->
00:15:22,350
we should have a session on that.
373
00:15:22,350 -->
00:15:26,270
So yesterday I held our second digital legacy network
374
00:15:26,270 -->
00:15:30,450
live session where we talked about setting up
375
00:15:30,450 -->
00:15:34,710
your emergency recovery contacts for Google,
376
00:15:34,710 -->
00:15:36,990
Facebook, Apple, et cetera.
377
00:15:36,990 -->
00:15:38,750
And what's really interesting about them
378
00:15:38,750 -->
00:15:41,430
is every one of them does it very differently.
379
00:15:41,430 -->
00:15:43,550
So you might just think you're gonna go into
380
00:15:43,550 -->
00:15:45,150
your account settings and say,
381
00:15:45,150 -->
00:15:47,230
look for emergency recovery contacts
382
00:15:47,230 -->
00:15:50,070
that I can set up in case I die.
383
00:15:50,070 -->
00:15:53,510
And none of them really work exactly that way.
384
00:15:53,510 -->
00:15:56,470
So we had a good discussion and I'll be posting that in,
385
00:15:56,470 -->
00:15:57,630
there's a LinkedIn group called
386
00:15:57,630 -->
00:16:00,390
the Digital Legacy Network, DLN I call it.
387
00:16:00,390 -->
00:16:03,430
And so I'll be putting the recording in that group
388
00:16:03,430 -->
00:16:05,790
but feel free to come and join into that group
389
00:16:05,790 -->
00:16:09,790
and make comments, ask questions, et cetera.
390
00:16:09,790 -->
00:16:12,710
Cause I think the whole idea of the AI stuff
391
00:16:12,750 -->
00:16:14,150
is way beyond the scope of the book
392
00:16:14,150 -->
00:16:16,150
that I've covered so far.
393
00:16:16,150 -->
00:16:18,310
It was trying to get to the basics of making sure
394
00:16:18,310 -->
00:16:21,710
that your key important stuff can be accessed
395
00:16:21,710 -->
00:16:23,310
when needed by people.
396
00:16:23,310 -->
00:16:26,830
But the AI aspect of it needs to be discussed
397
00:16:26,830 -->
00:16:29,190
and people are gonna be asking,
398
00:16:29,190 -->
00:16:32,190
okay, how do I deal with these things now?
399
00:16:32,190 -->
00:16:34,190
It's a great question because you're right,
400
00:16:34,190 -->
00:16:36,710
like these AI agents will just continue on
401
00:16:36,710 -->
00:16:39,550
and they may not even know you're dead.
402
00:16:39,550 -->
00:16:40,390
That's right.
403
00:16:40,390 -->
00:16:42,950
But it's interesting that Google's product
404
00:16:42,950 -->
00:16:45,270
for doing recovery when you die is basically,
405
00:16:45,270 -->
00:16:47,750
it's called the Inactive Account Manager.
406
00:16:47,750 -->
00:16:52,310
And because Google has so many properties or services,
407
00:16:52,310 -->
00:16:54,710
it can sort of tell, if you stop using YouTube
408
00:16:54,710 -->
00:16:59,030
and Gmail all at once, then after some period
409
00:16:59,030 -->
00:17:01,150
of inactivity, a minimum of three months,
410
00:17:01,150 -->
00:17:04,950
which is I think crazy, but it will turn
411
00:17:04,950 -->
00:17:06,870
over your account.
412
00:17:06,870 -->
00:17:09,150
It won't make a determination whether you're dead
413
00:17:09,190 -->
00:17:11,110
or not, just saying this is inactive,
414
00:17:11,110 -->
00:17:13,630
so it has to go to somebody else now.
415
00:17:13,630 -->
00:17:16,430
But that whole idea of being inactive,
416
00:17:16,430 -->
00:17:19,790
how do you determine that when it lets your agent
417
00:17:19,790 -->
00:17:21,670
do stuff for you?
418
00:17:21,670 -->
00:17:22,870
Yep, exactly.
419
00:17:22,870 -->
00:17:24,870
Well, we'll have links to all of that
420
00:17:24,870 -->
00:17:26,950
that Scott just described in the show notes.
421
00:17:26,950 -->
00:17:28,430
So definitely check it out.
422
00:17:28,430 -->
00:17:31,830
We've had a lot of our audience help out with the book.
423
00:17:31,830 -->
00:17:34,510
Yeah, awesome, thanks so much to the listeners
424
00:17:34,510 -->
00:17:35,710
who already helped, yeah.
425
00:17:35,710 -->
00:17:36,710
Yeah, we really appreciate it.
426
00:17:36,710 -->
00:17:37,990
So only another week or two,
427
00:17:37,990 -->
00:17:41,070
but I think to get to comments in before I lock up
428
00:17:41,070 -->
00:17:43,990
this version and send it to Kindle, so.
429
00:17:43,990 -->
00:17:47,270
All right, well, thank you all for listening.
430
00:17:47,270 -->
00:17:50,110
And until next time, stay safe, stay secure,
431
00:17:50,110 -->
00:17:51,550
and stay privates.
432
00:17:51,550 -->
00:17:53,310
Thank you for listening or watching.
433
00:17:53,310 -->
00:17:56,070
If you liked this episode, hit subscribe,
434
00:17:56,070 -->
00:17:57,950
share it with your friends and colleagues,
435
00:17:57,950 -->
00:18:01,230
or jump into our community at sharedsecurity.net
436
00:18:01,230 -->
00:18:03,990
slash supporter to keep the conversation going.
437
00:18:03,990 -->
00:18:06,150
Thanks again, and we'll see you next week
438
00:18:06,150 -->
00:18:08,790
for another episode of Shared Security.
