1 00:00:00,110 --> 00:00:03,680 S1: Whether you're starting or scaling your company's security program, demonstrating 2 00:00:03,710 --> 00:00:07,220 S1: top notch security practices and establishing trust is more important 3 00:00:07,220 --> 00:00:14,900 S1: than ever. Vanta automates compliance for Soc2, ISO 27,001 and more, 4 00:00:14,900 --> 00:00:19,520 S1: saving you time and money while helping you build customer trust. Plus, 5 00:00:19,520 --> 00:00:23,780 S1: you can streamline security reviews by automating questionnaires and demonstrating 6 00:00:23,780 --> 00:00:27,500 S1: your security posture with a customer facing trust center, all 7 00:00:27,500 --> 00:00:32,970 S1: powered by AI. Over 7000 global companies like Atlassian, Flow 8 00:00:32,970 --> 00:00:36,090 S1: Health and Quora use Vanta to manage risk and prove 9 00:00:36,090 --> 00:00:40,560 S1: security in real time. Get $1,000 off Vanta when you 10 00:00:40,560 --> 00:00:48,180 S1: go to Vanta comm slash unsupervised. That's vanta.com/supervised for $1,000 off. 11 00:00:49,979 --> 00:00:53,339 S1: Welcome to Unsupervised Learning, a security, AI and meaning focused 12 00:00:53,350 --> 00:00:56,230 S1: podcast that looks at how best to thrive as humans 13 00:00:56,230 --> 00:01:00,430 S1: in a post AI world. It combines original ideas, analysis, 14 00:01:00,430 --> 00:01:03,670 S1: and mental models to bring not just the news, but 15 00:01:03,670 --> 00:01:11,319 S1: why it matters and how to respond. All right, welcome 16 00:01:11,319 --> 00:01:14,230 S1: to unsupervised Learning. This is Daniel Miessler. I am in 17 00:01:14,230 --> 00:01:17,960 S1: a hotel room in Vegas recording because I have to. 18 00:01:17,959 --> 00:01:22,250 S1: But it is Hacker Week and I'm here for Defcon 19 00:01:22,250 --> 00:01:24,830 S1: and Black Hat and all that good stuff, but I 20 00:01:24,830 --> 00:01:28,459 S1: wanted to get the episode out. So first note here 21 00:01:28,459 --> 00:01:31,220 S1: is that Osint is one of my favorite hobbies, and 22 00:01:31,220 --> 00:01:33,950 S1: there's something called a pizza index. That's one of my 23 00:01:33,950 --> 00:01:37,970 S1: favorite examples of this, which is how much pizza essentially 24 00:01:37,970 --> 00:01:41,390 S1: the neighborhood around the Pentagon is ordering, which really means 25 00:01:41,390 --> 00:01:44,940 S1: the Pentagon. And there's another index related to that, which 26 00:01:44,940 --> 00:01:48,480 S1: is how many people are in the bars and this 27 00:01:48,480 --> 00:01:53,610 S1: person real. Ben Geller posted a tweet about this. And 28 00:01:53,610 --> 00:01:57,210 S1: essentially it says it shows that the people in the 29 00:01:57,210 --> 00:02:01,440 S1: bars is like extremely low and the pizza meter is 30 00:02:01,440 --> 00:02:04,560 S1: off the charts. And I just love this so much 31 00:02:04,560 --> 00:02:08,920 S1: because it indicates pretty strongly that something is about to 32 00:02:08,919 --> 00:02:10,870 S1: go down. And I've got a friend who used to 33 00:02:10,870 --> 00:02:13,870 S1: be an analyst at the Pentagon, and he says, this 34 00:02:13,870 --> 00:02:16,990 S1: is absolutely true. When people are ordering in pizza and 35 00:02:16,990 --> 00:02:20,619 S1: nobody's going home, it's obviously because something is going down. 36 00:02:20,620 --> 00:02:22,840 S1: And in this case, we kind of know what's going down, 37 00:02:22,840 --> 00:02:27,100 S1: which is Iran is preparing to attack Israel and or 38 00:02:27,100 --> 00:02:31,130 S1: whoever else. So that's what that is. But definitely check 39 00:02:31,130 --> 00:02:34,100 S1: out this tweet. It's pretty interesting. So this is also 40 00:02:34,100 --> 00:02:37,220 S1: why I can't wait to fully build out my agent framework, 41 00:02:37,220 --> 00:02:41,090 S1: and for agent framework to become more tightly integrated with 42 00:02:41,090 --> 00:02:44,030 S1: models and platforms, because it's going to allow a lot 43 00:02:44,030 --> 00:02:46,550 S1: more people to do things like this. What I love 44 00:02:46,550 --> 00:02:50,480 S1: about it is you could track all the different experts, right? 45 00:02:50,480 --> 00:02:52,250 S1: I'm going to use a whole bunch of different stuff 46 00:02:52,250 --> 00:02:55,200 S1: for this, but there will be some agent functionality in 47 00:02:55,200 --> 00:02:58,920 S1: the middle to sort of handle, like orchestration and summarization 48 00:02:58,919 --> 00:03:01,650 S1: and creating like an Intel report. But I love the 49 00:03:01,650 --> 00:03:08,130 S1: idea of like gathering all these individual, hopefully standalone intelligence sources, 50 00:03:08,130 --> 00:03:12,780 S1: aggregating them together, but also keeping them separate and then 51 00:03:12,780 --> 00:03:16,920 S1: triangulating on truth based on that. And I heard some 52 00:03:16,919 --> 00:03:20,320 S1: pretty cool ideas from, um, it was actually a friend 53 00:03:20,320 --> 00:03:23,709 S1: of mine named John, uh, who was talking about how 54 00:03:23,710 --> 00:03:26,350 S1: you want to rate those different sources in different ways. 55 00:03:26,350 --> 00:03:29,260 S1: One way to rate them is to rate them based 56 00:03:29,260 --> 00:03:33,280 S1: on their difference and their uniqueness of ideas relative to 57 00:03:33,280 --> 00:03:35,980 S1: other people, because you don't know if they're actually just 58 00:03:35,980 --> 00:03:39,670 S1: reading other people's Intel and following along, and you don't 59 00:03:39,670 --> 00:03:42,310 S1: want to use eight of those people who are all 60 00:03:42,310 --> 00:03:46,820 S1: following the same thing as eight different ends, right? Eight 61 00:03:46,820 --> 00:03:51,170 S1: different sources of of data or sources of signal. So 62 00:03:51,170 --> 00:03:53,300 S1: there's a whole bunch of cool stuff that you could 63 00:03:53,300 --> 00:03:56,480 S1: do once you have the discrete signal coming in from 64 00:03:56,480 --> 00:03:59,270 S1: all these different places. And then you could factor in 65 00:03:59,270 --> 00:04:02,960 S1: things like prediction markets and stuff like that and just, uh, 66 00:04:03,350 --> 00:04:05,840 S1: lots of different stuff you could do. But ultimately what 67 00:04:05,840 --> 00:04:09,570 S1: I want is a daily Intel report, which is as 68 00:04:09,570 --> 00:04:12,030 S1: good or better than what you would get from, like 69 00:04:12,030 --> 00:04:15,360 S1: Stratfor back in the day. Or you know what a 70 00:04:15,360 --> 00:04:19,230 S1: lot of these paid platforms would do. Um, or even like, 71 00:04:19,230 --> 00:04:21,869 S1: you know, a high level government thing, I think we 72 00:04:21,870 --> 00:04:26,760 S1: could build something really, really good that leverages the intelligence 73 00:04:26,760 --> 00:04:29,610 S1: of all these different, really smart people who are posting 74 00:04:29,610 --> 00:04:31,920 S1: their stuff online. We're not talking about private stuff. We're 75 00:04:31,930 --> 00:04:35,560 S1: talking about people on Twitter. We're talking about people on 76 00:04:35,560 --> 00:04:39,640 S1: different platforms, blogs. They're writing their stuff out there. And 77 00:04:39,640 --> 00:04:42,039 S1: a lot of times nobody's reading it. But if you 78 00:04:42,040 --> 00:04:44,200 S1: put the effort in, you could find all those signals 79 00:04:44,200 --> 00:04:48,490 S1: and start triangulating. So really excited about that. Okay. The 80 00:04:48,490 --> 00:04:52,330 S1: state of things. Yeah. I wrote a long piece about, uh, 81 00:04:52,330 --> 00:04:55,359 S1: this is I posted it on X, a fairly long piece. 82 00:04:55,360 --> 00:04:57,470 S1: I should probably turn it into a full blog, but 83 00:04:57,470 --> 00:05:01,310 S1: it's a little bit, uh, long winded and it's got 84 00:05:01,310 --> 00:05:02,900 S1: some politics in it, so I think I'm going to 85 00:05:02,900 --> 00:05:06,770 S1: skip it, but I recommend going and checking it out 86 00:05:06,770 --> 00:05:09,260 S1: if you're into that kind of stuff. And I spoke 87 00:05:09,260 --> 00:05:14,180 S1: with Christine Gadsby, the head of product security operations at BlackBerry, 88 00:05:14,180 --> 00:05:16,970 S1: and we talked about the role of AI in cybersecurity 89 00:05:16,970 --> 00:05:20,240 S1: and a whole bunch of different topics. The topic list 90 00:05:20,240 --> 00:05:23,970 S1: for this episode is quite large and you should absolutely 91 00:05:23,970 --> 00:05:26,640 S1: check it out. Um, so so go check that out 92 00:05:26,640 --> 00:05:28,679 S1: on the YouTube or you can click it in the 93 00:05:28,680 --> 00:05:32,609 S1: newsletter as well. So for security, two critical ServiceNow vulnerabilities 94 00:05:32,610 --> 00:05:36,870 S1: were reported by Asset Note company has reportedly paid a 95 00:05:36,870 --> 00:05:40,620 S1: new record high 75 million to a ransomware group. And 96 00:05:40,620 --> 00:05:42,839 S1: that seems like a lot of money. But it's not 97 00:05:42,839 --> 00:05:44,979 S1: a lot compared to not being able to do business 98 00:05:44,980 --> 00:05:47,110 S1: at all. So a lot of people kind of beat 99 00:05:47,110 --> 00:05:50,440 S1: up people for paying ransoms. And it really is kind 100 00:05:50,440 --> 00:05:52,990 S1: of similar to like, your kid gets stolen. It's like 101 00:05:52,990 --> 00:05:56,800 S1: all the philosophy goes away when somebody has your kid 102 00:05:56,800 --> 00:06:00,040 S1: and it's the same as a CEO or whoever. When 103 00:06:00,040 --> 00:06:01,960 S1: you have the ability to pay some money and get 104 00:06:01,960 --> 00:06:05,680 S1: business back online, sure, they might just ransom you again. Sure, 105 00:06:05,680 --> 00:06:08,470 S1: they might do whatever. Sure, it might be bad for 106 00:06:08,470 --> 00:06:12,830 S1: other people. But when business has stopped, things become quite 107 00:06:12,830 --> 00:06:15,830 S1: clear to you in terms of what you need to do. 108 00:06:15,830 --> 00:06:19,460 S1: So I'm not saying people should pay or anything like that. 109 00:06:19,460 --> 00:06:22,310 S1: I'm not making any judgments. I'm just saying, well, essentially 110 00:06:22,310 --> 00:06:26,030 S1: that don't make judgments. Try to avoid making judgments because 111 00:06:26,029 --> 00:06:28,580 S1: it's really hard to be in that position. Digicert is 112 00:06:28,580 --> 00:06:34,529 S1: revoking 83,000 TLS certificates due to a domain validation bug. 113 00:06:34,560 --> 00:06:37,830 S1: China is getting around US bans on advanced AI chips 114 00:06:37,830 --> 00:06:42,900 S1: through smuggling front companies and loopholes, basically finding ways to 115 00:06:42,900 --> 00:06:45,720 S1: get the chips that are not supposed to be getting. 116 00:06:45,750 --> 00:06:49,170 S1: Ransomware attacks are rising, with an 18% year on year 117 00:06:49,170 --> 00:06:54,089 S1: increase reported by Zscaler, and I've always considered ransomware attacks 118 00:06:54,089 --> 00:06:57,180 S1: to be something that we'd have to invent as a government. 119 00:06:57,180 --> 00:06:59,290 S1: It would have to be like a government service if 120 00:06:59,290 --> 00:07:01,720 S1: if it didn't exist in the marketplace, like as a 121 00:07:01,720 --> 00:07:04,690 S1: way to test for bad security. And maybe you give 122 00:07:04,690 --> 00:07:07,510 S1: like a fine or something if people keep having the mistake. 123 00:07:07,510 --> 00:07:10,390 S1: But my intuition was that after a number of years 124 00:07:10,390 --> 00:07:13,780 S1: that we'd get harder and harder because security would increase. 125 00:07:13,780 --> 00:07:16,989 S1: So if these attacks are still increasing, I wonder what 126 00:07:16,990 --> 00:07:20,380 S1: the reason is. Is it because attackers are moving to 127 00:07:20,380 --> 00:07:23,180 S1: like the more vulnerable targets, or are they just getting 128 00:07:23,180 --> 00:07:26,150 S1: better at finding the holes or something else? Or all 129 00:07:26,150 --> 00:07:28,429 S1: of the above? Probably all of the above. But if 130 00:07:28,430 --> 00:07:32,180 S1: somebody has more insight on why things aren't getting tighter 131 00:07:32,180 --> 00:07:34,790 S1: or see, that's the trick is it doesn't mean things 132 00:07:34,790 --> 00:07:37,310 S1: aren't getting harder. Just because the number of attacks are 133 00:07:37,310 --> 00:07:40,610 S1: going up doesn't mean things aren't getting harder. They might 134 00:07:40,610 --> 00:07:43,310 S1: just be getting better faster. Got a great analysis here 135 00:07:43,310 --> 00:07:47,730 S1: of securing secrets in AWS. So the blog post discussing 136 00:07:47,730 --> 00:07:51,420 S1: creating custom implants for evasion by building them in C 137 00:07:51,450 --> 00:07:55,770 S1: and this thing details server setup, client functionality, and testing 138 00:07:55,770 --> 00:07:59,160 S1: against security tools. The average cost of a data breach 139 00:07:59,160 --> 00:08:04,680 S1: jumped 10% to 4.88 million in 23. China is tightening 140 00:08:04,680 --> 00:08:09,270 S1: its civilian drone export rules starting September 1st to prevent 141 00:08:09,270 --> 00:08:13,040 S1: use in military or terrorist activities. Yeah, I'm trying to 142 00:08:13,040 --> 00:08:17,060 S1: figure out if this is CCP trying to keep it 143 00:08:17,060 --> 00:08:20,450 S1: their stuff from being used against them, or if they're 144 00:08:20,450 --> 00:08:25,220 S1: trying to make it easier to sell their products because 145 00:08:25,220 --> 00:08:29,330 S1: they're playing nice and they're appearing to be good guys. 146 00:08:29,360 --> 00:08:33,050 S1: AI and tech open AI has started rolling out its 147 00:08:33,050 --> 00:08:37,820 S1: new ChatGPT voice feature for ChatGPT plus users, and it's 148 00:08:37,820 --> 00:08:40,189 S1: quite good. It's it's quite a bit different. You can 149 00:08:40,190 --> 00:08:43,190 S1: basically interrupt it. It sounds a lot more natural. I 150 00:08:43,190 --> 00:08:45,860 S1: am getting a lot of voice artifacts though, like it'll 151 00:08:45,860 --> 00:08:49,310 S1: sound like choppy and broken and a lot of weird pauses. 152 00:08:49,309 --> 00:08:51,350 S1: Not like in a human way, but I think the 153 00:08:51,350 --> 00:08:54,260 S1: platform might be overwhelmed. Or maybe, I don't know, maybe 154 00:08:54,260 --> 00:08:56,510 S1: I need to restart the app. Maybe it was buggy. 155 00:08:56,510 --> 00:09:00,270 S1: Not sure. Lots of AI talk at Blackhat, which, uh, yeah, 156 00:09:00,270 --> 00:09:02,910 S1: already here and it's already happening. Um, another thing to 157 00:09:02,910 --> 00:09:06,330 S1: mention about the ChatGPT stuff is, uh, Greg Brockman is 158 00:09:06,330 --> 00:09:11,340 S1: taking a sabbatical. Uh, John Schulman, I think, is leaving 159 00:09:11,340 --> 00:09:14,640 S1: the company. Is he the one that went to to anthropic? 160 00:09:14,640 --> 00:09:17,969 S1: I can't remember. Another leader went to anthropic and another 161 00:09:17,970 --> 00:09:20,550 S1: one left as well. So the three people left all 162 00:09:20,550 --> 00:09:24,880 S1: at once. But it's not like a mass exodus all 163 00:09:24,880 --> 00:09:28,420 S1: to one place they're not mad at. OpenAI seems to 164 00:09:28,420 --> 00:09:31,450 S1: be fairly benign. Um, but it does look kind of 165 00:09:31,450 --> 00:09:33,640 S1: weird to have an announcement where three people leave at 166 00:09:33,640 --> 00:09:36,160 S1: the same time. A funniest joke I saw about this 167 00:09:36,160 --> 00:09:39,130 S1: was that Sam Altman predicted that soon there would be 168 00:09:39,130 --> 00:09:43,390 S1: a one person unicorn company, and the joke was, yeah, 169 00:09:43,390 --> 00:09:45,820 S1: it might be your company. You might be the only 170 00:09:45,820 --> 00:09:48,680 S1: one left. Um, I thought that was kind of clever. 171 00:09:48,679 --> 00:09:54,380 S1: California's SB 1047 safe and secure innovation for Frontier Artificial 172 00:09:54,380 --> 00:09:57,740 S1: Intelligence Models Act. That's a long name. It's looking to 173 00:09:57,740 --> 00:10:01,970 S1: regulate large AI models by mandating safety features to prevent 174 00:10:01,970 --> 00:10:08,660 S1: catastrophic incidents. Use risk based AI regulation began on August 1st, 175 00:10:08,660 --> 00:10:13,140 S1: and it's got staggered deadlines based on low or no 176 00:10:13,140 --> 00:10:17,160 S1: risk versus high risk and limited risk tiers. So that's 177 00:10:17,160 --> 00:10:20,370 S1: starting to roll out. And OpenAI has launched the GPT 178 00:10:20,370 --> 00:10:24,569 S1: four long output model. I've already switched all or at 179 00:10:24,570 --> 00:10:27,270 S1: least a lot of my stuff. I switched my fabric 180 00:10:27,270 --> 00:10:31,740 S1: prompt over to this. So it's got 64 output tokens, 181 00:10:31,740 --> 00:10:36,370 S1: 64,000 output tokens, which is 16 times more than the 182 00:10:36,370 --> 00:10:41,290 S1: previous one, and it's 50% cheaper for most things. And 183 00:10:41,290 --> 00:10:43,479 S1: a lot of people are saying that the benchmarks, it's 184 00:10:43,480 --> 00:10:46,420 S1: actually much better than the previous one. So I consider 185 00:10:46,420 --> 00:10:50,110 S1: it just a straight across upgrade plus being cheaper. So yeah, 186 00:10:50,110 --> 00:10:54,670 S1: I already made that change. Google's experimental Gemini 1.5 Pro 187 00:10:54,670 --> 00:10:58,780 S1: has claimed top spot on a bunch of leaderboards, surpassing 188 00:10:58,780 --> 00:11:03,740 S1: GPT four and, uh, sonnet 35 with a score of 1300. 189 00:11:03,770 --> 00:11:06,290 S1: I've not used it yet, because every time I try 190 00:11:06,290 --> 00:11:08,990 S1: to use a Google product, I have to vomit. But 191 00:11:08,990 --> 00:11:11,810 S1: I am going to try again soon to see, uh, 192 00:11:11,990 --> 00:11:14,810 S1: if it's usable. Meta says it'll need ten times more 193 00:11:14,809 --> 00:11:18,260 S1: computing power to train llama four compared to llama three. 194 00:11:18,290 --> 00:11:22,220 S1: Elliott Management is calling Nvidia a bubble and says AI 195 00:11:22,220 --> 00:11:25,650 S1: is overhyped. They mark. They argue that the market is 196 00:11:25,650 --> 00:11:29,880 S1: overly optimistic about AI's potential and Nvidia's role in it. 197 00:11:29,880 --> 00:11:31,980 S1: I think it's a bubble, but it's a bubble like 198 00:11:31,980 --> 00:11:36,059 S1: the internet in 1995. In other words, there absolutely will 199 00:11:36,059 --> 00:11:39,030 S1: be a burst of lots and lots of companies, right? 200 00:11:39,030 --> 00:11:44,160 S1: Pets.com and companies like that, the AI equivalents, thousands of 201 00:11:44,160 --> 00:11:47,280 S1: those companies are going to fail. Lots of investors are 202 00:11:47,280 --> 00:11:49,630 S1: going to be very sad about this, but that's completely 203 00:11:49,630 --> 00:11:52,930 S1: unrelated to what AI is up is about to do 204 00:11:52,929 --> 00:11:55,270 S1: to the world. Right? So I think people shouldn't be 205 00:11:55,270 --> 00:11:57,970 S1: confused about those two things. One happening doesn't mean that 206 00:11:57,970 --> 00:12:00,490 S1: the other one is not going to happen. Bellingcat has 207 00:12:00,490 --> 00:12:05,380 S1: put together a guide on identifying explosive ordnance in social 208 00:12:05,380 --> 00:12:09,850 S1: media imagery. CrowdStrike is facing a massive lawsuit after Blue 209 00:12:09,850 --> 00:12:13,610 S1: Friday crashed over 8 million computers globally. Intel is laying 210 00:12:13,610 --> 00:12:16,220 S1: off over 15% of its workforce as part of a 211 00:12:16,220 --> 00:12:20,059 S1: $10 billion cost reduction plan. Apple just posted a record 212 00:12:20,059 --> 00:12:25,219 S1: breaking Q3 2024, $86 billion in revenue. And one thing 213 00:12:25,220 --> 00:12:28,729 S1: that's interesting about this is Berkshire Hathaway just sold a 214 00:12:28,730 --> 00:12:31,130 S1: whole bunch of stuff, uh, a whole bunch of Apple. 215 00:12:31,130 --> 00:12:34,370 S1: And they sold it right before this crash happened. The 216 00:12:34,370 --> 00:12:37,220 S1: crash happened. There was a giant recession that hit the 217 00:12:37,220 --> 00:12:39,380 S1: United States, and then it went away the next day. 218 00:12:39,410 --> 00:12:45,050 S1: Today was like a lot of that money came back. But, um. Yeah, strange. 219 00:12:45,050 --> 00:12:48,740 S1: Who knows? It could happen again tomorrow. But very volatile, 220 00:12:48,740 --> 00:12:52,130 S1: very emotional sort of time. I feel like in lots 221 00:12:52,130 --> 00:12:54,440 S1: of different ways, and I feel like the stock market 222 00:12:54,440 --> 00:12:57,500 S1: is matching that. But, uh, the other thing to mention 223 00:12:57,500 --> 00:13:02,250 S1: about Apple is that their services money is now almost 224 00:13:02,250 --> 00:13:05,910 S1: equal to their devices money, which is a huge tipping 225 00:13:05,910 --> 00:13:09,959 S1: point or a milestone in terms of their growth. Apple 226 00:13:09,960 --> 00:13:13,199 S1: is ramping up spending to get Apple intelligence ready for 227 00:13:13,200 --> 00:13:16,830 S1: launch in the fall. I'm already using the beta, and 228 00:13:16,830 --> 00:13:19,080 S1: it's pretty impressive, even though a lot of the features 229 00:13:19,080 --> 00:13:22,170 S1: aren't rolled out yet. All right, human news. A lot 230 00:13:22,170 --> 00:13:25,140 S1: of the world tried to push Huawei out of their infrastructure, 231 00:13:25,140 --> 00:13:30,250 S1: but they're actually getting more successful, not less. Software company 232 00:13:30,250 --> 00:13:35,140 S1: increased user engagement by eight times by drastically shortening their emails. 233 00:13:35,140 --> 00:13:40,059 S1: Netlify fees. Is that it? Yeah. Netlify fees. Initial 150 234 00:13:40,059 --> 00:13:43,720 S1: word emails had a 1% reply rate, but by cutting 235 00:13:43,720 --> 00:13:47,110 S1: the text to 37 words, it went to 4%. And 236 00:13:47,110 --> 00:13:50,660 S1: when they went to 14 words, it went to 8% 237 00:13:50,660 --> 00:13:54,470 S1: 14 words. Last month, Shane Mack offered everyone at his 238 00:13:54,470 --> 00:13:58,160 S1: company $25,000 to quit and six people took it. Yeah, 239 00:13:58,160 --> 00:14:01,160 S1: I think this is part of the Alaskan fishing boat 240 00:14:01,160 --> 00:14:03,740 S1: thing that I wrote a while back. Companies basically want 241 00:14:03,740 --> 00:14:07,460 S1: fully dedicated murderers is all they want. They want people 242 00:14:07,460 --> 00:14:11,330 S1: who eat, live, sleep, think and are obsessed with the company. 243 00:14:11,330 --> 00:14:15,270 S1: That's why they want return to office. That's their way 244 00:14:15,270 --> 00:14:18,210 S1: of filtering for people who who think of the company 245 00:14:18,210 --> 00:14:20,700 S1: as a religion. I mean, they can't say that, but 246 00:14:20,700 --> 00:14:22,229 S1: they can say you have to come to the office 247 00:14:22,230 --> 00:14:25,500 S1: and that's an automatic filter for it. Right? So this 248 00:14:25,500 --> 00:14:28,020 S1: is the way that management and managers and the whole 249 00:14:28,020 --> 00:14:32,490 S1: whole system can basically look for these obsessed people, which 250 00:14:32,490 --> 00:14:36,420 S1: are likely to be in certain demographics, right? Certain ages, 251 00:14:36,420 --> 00:14:40,930 S1: you know, certain groups that are awfully likely to look 252 00:14:40,930 --> 00:14:44,620 S1: kind of similar to each other. Probably young, probably without kids, 253 00:14:44,620 --> 00:14:48,040 S1: probably male, who are just grind, grind, grind, don't care 254 00:14:48,040 --> 00:14:51,520 S1: about anything else. Yeah, whatever. Work life balance don't care. 255 00:14:51,520 --> 00:14:54,040 S1: I just want a code or whatever it is. Right. 256 00:14:54,040 --> 00:14:57,160 S1: So that's what these companies are looking for more and more. 257 00:14:57,160 --> 00:14:59,890 S1: And that's why I think and this is just my 258 00:14:59,890 --> 00:15:03,380 S1: hypothesis here, I don't, you know, we need more data 259 00:15:03,380 --> 00:15:07,940 S1: for all this. But my pet hypothesis here is that 260 00:15:07,940 --> 00:15:11,000 S1: this is a factor for all of these layoffs. It's 261 00:15:11,000 --> 00:15:15,590 S1: like this awakening across all of business that you know 262 00:15:15,590 --> 00:15:21,170 S1: what I want hardcore crazy people, religious people about this company. 263 00:15:21,170 --> 00:15:23,660 S1: And I want them to be a-players. And I want 264 00:15:23,660 --> 00:15:26,720 S1: them to be really good at AI, and they're going 265 00:15:26,720 --> 00:15:28,950 S1: to help us do even more with AI because they're 266 00:15:28,950 --> 00:15:31,080 S1: going to bring the AI on and blah, blah, blah. 267 00:15:31,080 --> 00:15:33,330 S1: So it's like, I'm going to hire a bunch of 268 00:15:33,330 --> 00:15:36,570 S1: these crazy people, and a team of ten of them 269 00:15:36,570 --> 00:15:39,030 S1: is going to be like having a team of 1000 270 00:15:39,030 --> 00:15:43,080 S1: or 2000 people sometime in the future, in the near future. 271 00:15:43,080 --> 00:15:44,880 S1: Whereas if you get a bunch of people who are 272 00:15:44,880 --> 00:15:48,270 S1: just straight out of college, they're entitled. They think they 273 00:15:48,270 --> 00:15:51,210 S1: are owed something even worse. They think that they're about 274 00:15:51,210 --> 00:15:54,550 S1: to receive training on the job because they don't know 275 00:15:54,550 --> 00:15:57,010 S1: how to do the job. And it's like, okay, well, 276 00:15:57,010 --> 00:15:59,470 S1: now train me. Now, teach me how to do this job. 277 00:15:59,470 --> 00:16:02,500 S1: And all these leaders at these companies are like, I 278 00:16:02,500 --> 00:16:05,350 S1: do not want you. I don't care what degrees you have. 279 00:16:05,350 --> 00:16:07,810 S1: If you can't do the job on day one, or 280 00:16:07,810 --> 00:16:11,440 S1: you can't learn instantly, like just by seeing it once, 281 00:16:11,440 --> 00:16:13,150 S1: and if you're not obsessed about it and want to 282 00:16:13,150 --> 00:16:16,150 S1: sleep under your desk, we have no use for you. 283 00:16:16,160 --> 00:16:20,510 S1: And unfortunately, that's like 80% of the workforce, I'm guessing, right? 284 00:16:20,510 --> 00:16:24,410 S1: It's like 52, 90% of the workforce, let's call it that. 285 00:16:24,410 --> 00:16:27,920 S1: And what that means is they are looking for that 10%. 286 00:16:27,920 --> 00:16:31,400 S1: They're looking for that 5%. They're looking for the A 287 00:16:31,400 --> 00:16:36,320 S1: players who are dedicated like religious people. And I believe 288 00:16:36,320 --> 00:16:39,620 S1: this is what we're seeing more than anything now. You 289 00:16:39,620 --> 00:16:41,910 S1: add AI on top of that. Now you see why 290 00:16:41,910 --> 00:16:43,920 S1: there's so many layoffs. Now you see why there's so 291 00:16:43,920 --> 00:16:47,160 S1: many open positions. But nobody's hiring for them because they're 292 00:16:47,160 --> 00:16:51,060 S1: kind of like fake positions. And this is multiple hypotheses 293 00:16:51,060 --> 00:16:54,210 S1: all rolled into one. But you get the vibe. This 294 00:16:54,210 --> 00:16:56,880 S1: is the basic vibe of what I think is happening. 295 00:16:56,880 --> 00:17:01,980 S1: Journalist Evan Gershkovich was among a group of Americans and 296 00:17:01,980 --> 00:17:06,550 S1: Russian dissidents released from Russia in a seven nation prisoner swap, 297 00:17:06,550 --> 00:17:10,389 S1: largest ever since the Cold War. Researchers at the University 298 00:17:10,390 --> 00:17:13,750 S1: of California, Santa Barbara have developed an AI model called 299 00:17:13,750 --> 00:17:17,080 S1: shark AI to help prevent shark attacks. The model uses 300 00:17:17,080 --> 00:17:21,369 S1: drones to detect sharks with greater accuracy than humans. I 301 00:17:21,369 --> 00:17:24,550 S1: love this, I love this every time I go to Maui, 302 00:17:24,550 --> 00:17:27,310 S1: I'm stupid and I read the stats about like, shark 303 00:17:27,310 --> 00:17:29,780 S1: attacks and they're like, oh, actually, right next to you 304 00:17:29,780 --> 00:17:32,149 S1: is the most dangerous place. And I'm like, cool. I 305 00:17:32,150 --> 00:17:34,130 S1: didn't want to go in the water anyway. Why did 306 00:17:34,130 --> 00:17:37,040 S1: I read that right before I went on vacation, where 307 00:17:37,040 --> 00:17:39,740 S1: I'm supposed to swim in the water? But anyway, if 308 00:17:39,740 --> 00:17:42,170 S1: I were able to look up and maybe they're so 309 00:17:42,170 --> 00:17:45,050 S1: high up you can't hear them, maybe it's not super annoying. 310 00:17:45,050 --> 00:17:47,750 S1: But anyway, if I know that there's ten of these 311 00:17:47,750 --> 00:17:53,000 S1: drones sweeping back and forth and you know they're being recharged, 312 00:17:53,000 --> 00:17:55,530 S1: they go back on rotation and they look down. They 313 00:17:55,530 --> 00:17:58,470 S1: could see very clearly if there's a shark in the water, 314 00:17:58,470 --> 00:18:01,050 S1: I assume that it wouldn't work. Maybe if the water 315 00:18:01,050 --> 00:18:04,380 S1: was muddy, but maybe you wouldn't be swimming anyway because 316 00:18:04,380 --> 00:18:07,380 S1: it would be dangerous water anyway. Usually in a lot 317 00:18:07,380 --> 00:18:09,659 S1: of places you could see right through the water. It's 318 00:18:09,660 --> 00:18:13,560 S1: very easy to see a shark from above, and they 319 00:18:13,560 --> 00:18:18,520 S1: just call the lifeguard station and trigger an alert, and 320 00:18:18,520 --> 00:18:20,980 S1: they blow the whistle and everyone gets out of the water, like, 321 00:18:20,980 --> 00:18:24,130 S1: that's going to be amazing. Love it. Treating failing eyesight 322 00:18:24,130 --> 00:18:26,410 S1: and high cholesterol are two new ways to lower the 323 00:18:26,410 --> 00:18:30,130 S1: risk of developing dementia, according to a major report. The 324 00:18:30,130 --> 00:18:35,500 S1: Lancet Commission's latest findings suggest that addressing 14 health issues 325 00:18:35,500 --> 00:18:39,820 S1: could theoretically prevent nearly half of all dementia cases worldwide. 326 00:18:39,820 --> 00:18:43,040 S1: And I believe from reading this that essentially they're talking 327 00:18:43,040 --> 00:18:46,550 S1: about things that just exacerbate it and make it worse. So, 328 00:18:46,550 --> 00:18:49,640 S1: for example, if you can't really see things, you generally 329 00:18:49,640 --> 00:18:52,040 S1: maybe you don't go out a lot. If you can't 330 00:18:52,040 --> 00:18:55,010 S1: hear conversations, you're not involved in conversations. So I think 331 00:18:55,010 --> 00:18:58,130 S1: a lot of this might be related to social interaction, 332 00:18:58,130 --> 00:19:01,250 S1: which once you start to get isolated and you're not 333 00:19:01,250 --> 00:19:04,400 S1: consuming media, you're not reading, you're not like, there's no 334 00:19:04,400 --> 00:19:07,830 S1: new inputs. Um, again, this is my hypothesis. I believe 335 00:19:07,830 --> 00:19:10,109 S1: this is based on some solid science I've already read, 336 00:19:10,109 --> 00:19:13,770 S1: though is basically, once you get isolated in that way, 337 00:19:13,770 --> 00:19:17,340 S1: your brain starts like shutting down, and it really accelerates 338 00:19:17,340 --> 00:19:21,180 S1: the dementia. So that that would make sense if, uh, 339 00:19:21,180 --> 00:19:23,669 S1: that's what they were saying in this paper. Self control 340 00:19:23,670 --> 00:19:27,990 S1: is about 60% heritable, meaning genes explain roughly 60% of 341 00:19:27,990 --> 00:19:32,170 S1: the differences in self control among individuals. So I think 342 00:19:32,170 --> 00:19:34,930 S1: this could be devastating if it's supported in further studies. 343 00:19:34,930 --> 00:19:39,159 S1: I worry about the narrative that both IQ and self-discipline 344 00:19:39,160 --> 00:19:43,000 S1: are mostly genetic, thus giving people an easy ramp to 345 00:19:43,000 --> 00:19:46,629 S1: write off individuals or even groups if they have lower 346 00:19:46,630 --> 00:19:50,950 S1: averages of these things. And I think even if it 347 00:19:50,950 --> 00:19:55,400 S1: were true, the groups don't define the individuals and the 348 00:19:55,400 --> 00:19:58,550 S1: study mentioned individuals here. It's not talking about groups, but 349 00:19:58,550 --> 00:20:01,159 S1: you know, people are going to people. Right? So the 350 00:20:01,160 --> 00:20:05,210 S1: the other thing is there's likely a lot of slack in, say, 351 00:20:05,210 --> 00:20:09,260 S1: the 40%, which is environmental, assuming those numbers are correct, 352 00:20:09,260 --> 00:20:13,490 S1: like we're probably getting whatever, 10 or 20% of the 40% 353 00:20:13,490 --> 00:20:17,240 S1: we're supposed to be doing. So if we were to increase, 354 00:20:17,240 --> 00:20:20,520 S1: you know, the efforts of, you know, training and culture 355 00:20:20,520 --> 00:20:23,639 S1: and all the environmental things we can control, I think 356 00:20:23,640 --> 00:20:27,570 S1: that would raise, you know, the bar for, well, everyone, 357 00:20:27,570 --> 00:20:30,600 S1: but especially the bottom quite a bit. So I'm not 358 00:20:30,600 --> 00:20:33,570 S1: sure this is really anything too much to despair about 359 00:20:33,570 --> 00:20:37,980 S1: other than making it easier for certain negative narratives. A 360 00:20:37,980 --> 00:20:40,620 S1: new study reveals that people tend to alter their appearance 361 00:20:40,619 --> 00:20:44,899 S1: to match their names. Researchers found that adults faces often 362 00:20:44,900 --> 00:20:49,400 S1: align with a social stereotype associated with their name, while 363 00:20:49,400 --> 00:20:52,790 S1: children's faces do not show this pattern. A key protein 364 00:20:52,790 --> 00:20:56,720 S1: called reelin may help stave off Alzheimer's disease. A number 365 00:20:56,720 --> 00:20:59,720 S1: of new studies suggest that reelin helps maintain thinking and 366 00:20:59,720 --> 00:21:04,639 S1: memory in aging brains, and when its levels fall off, 367 00:21:04,640 --> 00:21:08,689 S1: neurons become more vulnerable and people are starting to obviously 368 00:21:08,690 --> 00:21:11,780 S1: work on drugs for this. Wizards of the coast will 369 00:21:11,780 --> 00:21:16,910 S1: release the 2024 Dungeons and Dragons rulebooks under a Creative 370 00:21:16,910 --> 00:21:21,200 S1: Commons license, which is fulfilling a promise they made after 371 00:21:21,200 --> 00:21:24,440 S1: the backlash over attempts to change the Open Gaming License. 372 00:21:24,440 --> 00:21:27,950 S1: If novelists wrote Your bug Reports imagines how famous authors 373 00:21:27,950 --> 00:21:32,100 S1: would describe software bugs in their unique styles, Ernest Klein 374 00:21:32,100 --> 00:21:35,100 S1: likens a screen flicker to scenes from back to the 375 00:21:35,100 --> 00:21:40,590 S1: Future and Ghostbusters, while Ursula K Le Guin philosophizes about 376 00:21:40,590 --> 00:21:45,540 S1: the existential pain of coding errors. Ideas. More analysis on 377 00:21:45,540 --> 00:21:48,870 S1: how bad the results were of the recent UBI study 378 00:21:48,869 --> 00:21:52,950 S1: done by Sam Altman. It looks pretty bad, just like 379 00:21:52,950 --> 00:21:55,859 S1: we talked about last week, and got a link here 380 00:21:55,869 --> 00:21:58,629 S1: to go into that in depth. And really cool idea 381 00:21:58,630 --> 00:22:02,950 S1: from Jonathan Hite about free range kids. And a cool 382 00:22:02,950 --> 00:22:06,400 S1: idea for giving them freedom is to create a play 383 00:22:06,400 --> 00:22:08,919 S1: street once a month where you close off a street 384 00:22:08,920 --> 00:22:12,550 S1: for two hours, give time for kids to play in 385 00:22:12,550 --> 00:22:15,699 S1: the street safely, and then the whole time that the 386 00:22:15,700 --> 00:22:20,170 S1: parents are there watching, like around the edges and, you know, whatever. 387 00:22:20,180 --> 00:22:23,060 S1: But the neighbors are also meeting and talking, and he's 388 00:22:23,060 --> 00:22:27,320 S1: saying it has transformative effects on the neighborhood and just 389 00:22:27,320 --> 00:22:30,470 S1: good times all around. I really love ideas like this. 390 00:22:30,470 --> 00:22:34,100 S1: Discovery Farmbot is an open source farming machine for growing 391 00:22:34,130 --> 00:22:37,250 S1: food in your own backyard. Super memory an AI powered 392 00:22:37,250 --> 00:22:41,600 S1: platform to organize, search and utilize saved information acting as 393 00:22:41,600 --> 00:22:47,190 S1: a digital second brain. Friend is Avi Schiffman's new AI pendant, 394 00:22:47,190 --> 00:22:51,030 S1: and it's designed to combat loneliness by sending you reassuring 395 00:22:51,030 --> 00:22:56,310 S1: or playful text based on what it overhears so it 396 00:22:56,310 --> 00:22:59,730 S1: doesn't have a speaker. It actually sends you notifications. Kind 397 00:22:59,730 --> 00:23:02,760 S1: of interesting way to do that. Daniel Kosman walks you 398 00:23:02,760 --> 00:23:07,830 S1: through installing fabric and open source AI framework by Daniel Missler. 399 00:23:07,859 --> 00:23:10,540 S1: That's weird. I wonder if I wrote that because I 400 00:23:10,540 --> 00:23:13,060 S1: don't talk about myself in the third person. Fleet is 401 00:23:13,060 --> 00:23:15,970 S1: an open source version of fleet DMs tool built on 402 00:23:15,970 --> 00:23:22,270 S1: OS query for vulnerability monitoring, MDM detection engineering and more. 403 00:23:22,270 --> 00:23:27,460 S1: Soc2 policy Templates collection of templates for Soc2 policies and procedures. 404 00:23:27,460 --> 00:23:32,020 S1: Clutch security is a platform providing visibility into all non-human 405 00:23:32,020 --> 00:23:38,610 S1: identities within an organization, helping them identify associated risks and 406 00:23:38,609 --> 00:23:41,250 S1: the recommendation of the week. If you're at Blackhat this week, 407 00:23:41,250 --> 00:23:44,160 S1: remember that ten and 20 years from now, you will 408 00:23:44,160 --> 00:23:47,070 S1: not remember the talks that you saw this year, but 409 00:23:47,070 --> 00:23:50,970 S1: you will remember spending that time with your friends. So 410 00:23:50,970 --> 00:23:55,590 S1: prioritize friend time over presentation time. Not only is the 411 00:23:55,590 --> 00:23:58,620 S1: friend time more precious and valuable, but you can get 412 00:23:58,619 --> 00:24:00,959 S1: the talks later if you really want to. And the 413 00:24:00,960 --> 00:24:03,570 S1: aphorism of the week friends show their love in times 414 00:24:03,570 --> 00:24:07,320 S1: of trouble, not in happiness. Friends show their love in 415 00:24:07,320 --> 00:24:14,760 S1: times of trouble, not in happiness. Euripides. Unsupervised learning is 416 00:24:14,760 --> 00:24:17,880 S1: produced and edited by Daniel Miller on a Neumann U87 417 00:24:17,880 --> 00:24:21,970 S1: AI microphone using Hindenburg. Intro and outro music is by 418 00:24:21,970 --> 00:24:25,179 S1: Zomby with the Y, and to get the text and 419 00:24:25,180 --> 00:24:27,550 S1: links from this episode, sign up for the newsletter version 420 00:24:27,550 --> 00:24:33,190 S1: of the show at Daniel miessler.com/newsletter. We'll see you next time.