1 00:00:00,080 --> 00:00:02,540 S1: When you go through airport security, there's a line where 2 00:00:02,540 --> 00:00:05,600 S1: the TSA agent checks your ID, and another line where 3 00:00:05,600 --> 00:00:08,030 S1: a machine scans your bag. The same thing happens in 4 00:00:08,030 --> 00:00:11,540 S1: enterprise security, but instead of passengers and luggage, it's end 5 00:00:11,539 --> 00:00:14,720 S1: users and their devices. These days, most companies are pretty 6 00:00:14,720 --> 00:00:16,700 S1: good at the first part of the equation where they 7 00:00:16,700 --> 00:00:19,790 S1: check user identity, but user devices can roll right through 8 00:00:19,790 --> 00:00:23,780 S1: authentication without getting inspected at all. In fact, 47% of 9 00:00:23,780 --> 00:00:28,190 S1: companies allow unmanaged, untrusted devices to access their data. That 10 00:00:28,190 --> 00:00:30,320 S1: means an employee can log in from a laptop that 11 00:00:30,320 --> 00:00:33,320 S1: has its firewall turned off and hasn't been updated in 12 00:00:33,320 --> 00:00:36,380 S1: six months. Or worse, that laptop might be a bad 13 00:00:36,380 --> 00:00:40,519 S1: actor using employee credentials. One password finally solves the device 14 00:00:40,520 --> 00:00:43,700 S1: trust problem. One password ensures that no device can log 15 00:00:43,700 --> 00:00:47,839 S1: into your Okta protected apps unless it passes your security checks. Plus, 16 00:00:47,840 --> 00:00:51,409 S1: you can use one password on devices without MDM, like 17 00:00:51,409 --> 00:00:55,760 S1: your Linux fleet, contractor devices, and every BYoD phone and 18 00:00:55,760 --> 00:00:59,600 S1: laptop in your company. Visit one password comm slash unsupervised 19 00:00:59,600 --> 00:01:02,270 S1: learning to watch a demo and see how it works. 20 00:01:02,270 --> 00:01:12,140 S1: That's one password.com/unsupervised learning. Welcome to Unsupervised Learning, a security 21 00:01:12,170 --> 00:01:14,869 S1: I and meaning focused podcast that looks at how best 22 00:01:14,870 --> 00:01:18,230 S1: to thrive as humans in a post AI world. It 23 00:01:18,230 --> 00:01:22,250 S1: combines original ideas, analysis, and mental models to bring not 24 00:01:22,250 --> 00:01:25,369 S1: just the news, but why it matters and how to respond. 25 00:01:29,540 --> 00:01:32,660 S1: All right. Welcome to unsupervised learning. This is Daniel Meisler 26 00:01:32,660 --> 00:01:36,500 S1: and this is episode 431. This is RSA week. Got 27 00:01:36,500 --> 00:01:38,990 S1: grok support coming to fabric. If you've not messed with 28 00:01:38,990 --> 00:01:41,959 S1: grok yet you absolutely need to go do this. It 29 00:01:41,959 --> 00:01:46,520 S1: is insanely cool to use. It is so fast! I 30 00:01:46,520 --> 00:01:48,260 S1: would love to do a demo, but we need to 31 00:01:48,260 --> 00:01:51,320 S1: get through the show, so I'll probably do a demo 32 00:01:51,320 --> 00:01:55,250 S1: in a separate talk or uh, little video soon. My 33 00:01:55,250 --> 00:01:59,480 S1: buddy Clint Gibler and also Caleb Simon gave awesome talks 34 00:01:59,480 --> 00:02:02,300 S1: at Bsides, so that's super cool. Be sure to check 35 00:02:02,300 --> 00:02:04,309 S1: those out when they come out. I got a new 36 00:02:04,310 --> 00:02:07,880 S1: essay here on how consultancies are about to move into 37 00:02:07,880 --> 00:02:11,600 S1: departments and companies, and basically break them into pieces and 38 00:02:11,600 --> 00:02:14,000 S1: apply AI to them. I've been wanting to write this 39 00:02:14,000 --> 00:02:16,370 S1: one for a very long time. It's called companies are 40 00:02:16,370 --> 00:02:19,550 S1: a just a graph of algorithms, and I'm probably going 41 00:02:19,550 --> 00:02:22,549 S1: to record this as a standalone episode as well. Second 42 00:02:22,550 --> 00:02:25,730 S1: new essay this week is how I think prompting is 43 00:02:25,730 --> 00:02:28,820 S1: kind of the center of AI, and even though there's 44 00:02:28,820 --> 00:02:31,280 S1: lots of cool stuff you could do with like fine 45 00:02:31,280 --> 00:02:33,830 S1: tuning and training your own models and stuff, I think 46 00:02:33,830 --> 00:02:36,500 S1: prompting is still where it's at, and this is kind 47 00:02:36,500 --> 00:02:39,650 S1: of like an important walk through, or at least a 48 00:02:39,650 --> 00:02:44,210 S1: decent walk through. On why I believe that, security wise, 49 00:02:44,210 --> 00:02:47,720 S1: Biden administration is changing it so that you can they 50 00:02:47,720 --> 00:02:51,830 S1: can basically hire within it, and especially security. They can 51 00:02:51,830 --> 00:02:55,130 S1: hire essentially people who don't have a degree, which has 52 00:02:55,130 --> 00:02:58,070 S1: been a huge limiting factor for them to get talent. 53 00:02:58,070 --> 00:03:01,970 S1: So that's really, really good news. The CEO of UnitedHealth 54 00:03:01,970 --> 00:03:06,050 S1: took personal responsibility for paying the $22 million ransom to 55 00:03:06,050 --> 00:03:09,950 S1: get business back running, and that is really interesting that 56 00:03:09,950 --> 00:03:12,920 S1: he did that. We'll see what the fallout is. I 57 00:03:12,919 --> 00:03:15,170 S1: do worry a little bit about the signaling that it's 58 00:03:15,169 --> 00:03:17,600 S1: basically saying, hey, it's okay to pay ransoms, which of 59 00:03:17,600 --> 00:03:20,180 S1: course kind of propagates it. It's like, oh, we never 60 00:03:20,180 --> 00:03:23,780 S1: negotiate with terrorists type of deal. That policy makes it 61 00:03:23,780 --> 00:03:27,350 S1: so that people are less likely to become terrorists. So 62 00:03:27,350 --> 00:03:30,619 S1: we'll see how that plays out. Satya Nadella sent out 63 00:03:30,620 --> 00:03:33,770 S1: a Bill gates type memo saying that security was the 64 00:03:33,770 --> 00:03:36,890 S1: top priority, which is cool to see history repeat itself there. 65 00:03:36,890 --> 00:03:41,660 S1: Cybersecurity consultancy got busted for trying to well, no, a 66 00:03:41,660 --> 00:03:45,170 S1: consultant got busted for trying to extort an IT firm 67 00:03:45,170 --> 00:03:50,270 S1: for $1.5 million by threatening to leak their secrets. Not 68 00:03:50,270 --> 00:03:53,300 S1: the way to get what you want. Or maybe it is, 69 00:03:53,300 --> 00:03:56,750 S1: I don't know. Verizon, AT&T, T-Mobile and sprint got hit 70 00:03:56,750 --> 00:04:01,550 S1: with a $200 million fine for selling customer location data. Oh, 71 00:04:01,550 --> 00:04:04,820 S1: this is insane. A team trained a robot dog in 72 00:04:04,820 --> 00:04:08,810 S1: a simulation. Completely in a simulation how to walk on 73 00:04:08,810 --> 00:04:11,330 S1: a ball. And then they took that code and put 74 00:04:11,330 --> 00:04:13,580 S1: it in an actual robot dog and put it on 75 00:04:13,580 --> 00:04:17,029 S1: an actual ball. And it just worked. I mean, think 76 00:04:17,029 --> 00:04:19,430 S1: about how that's going to transfer to lots of different 77 00:04:19,430 --> 00:04:24,710 S1: human problems. Google is pushing for a change to immigration policies. 78 00:04:24,710 --> 00:04:27,710 S1: They basically say that we're losing AI and security talent 79 00:04:27,710 --> 00:04:30,800 S1: because they're going to other countries. This is a tweet 80 00:04:30,800 --> 00:04:34,010 S1: of mine about the difference between a company that uses 81 00:04:34,010 --> 00:04:37,969 S1: back end models versus the whole company is actually the 82 00:04:37,970 --> 00:04:41,179 S1: back end model. So it's like, what's your vulnerability to 83 00:04:41,180 --> 00:04:46,219 S1: getting sherlocked by someone like OpenAI or anthropic? And this 84 00:04:46,220 --> 00:04:49,880 S1: is basically a breakdown for how you should actually build 85 00:04:49,880 --> 00:04:52,159 S1: your company so that it just gets better when those 86 00:04:52,160 --> 00:04:55,280 S1: things improve, it doesn't actually get replaced. I got a 87 00:04:55,279 --> 00:04:58,310 S1: doctor buddy who loves AI note taking. She said she 88 00:04:58,310 --> 00:05:01,310 S1: was basically going to give up and just get out 89 00:05:01,310 --> 00:05:05,420 S1: of practicing medicine. And basically I note taking saved her 90 00:05:05,420 --> 00:05:09,170 S1: from doing that. So that's cool. Someone's criticizing Sam Altman's 91 00:05:09,170 --> 00:05:12,740 S1: approach as a blend of fear, ignoring uncertainties and riding 92 00:05:12,740 --> 00:05:16,820 S1: the hype wave. Apple is supposedly working on a big 93 00:05:16,820 --> 00:05:20,510 S1: AI team, pulling people from Google and building up some 94 00:05:20,510 --> 00:05:23,900 S1: sort of lab in Zurich. My buddy Joseph Thacker just 95 00:05:23,900 --> 00:05:28,550 S1: wrote a great post on assumptions in Lm's assumptions made. 96 00:05:28,550 --> 00:05:31,520 S1: This is the article you want to go check it out? It's, uh, 97 00:05:31,520 --> 00:05:36,289 S1: really good. Somebody automated a YouTube shorts channel entirely entirely 98 00:05:36,290 --> 00:05:40,880 S1: with free tools. So it's a short book. Summaries is 99 00:05:40,880 --> 00:05:43,820 S1: what it is, and it's just releasing episodes. It's doing 100 00:05:43,820 --> 00:05:47,089 S1: the whole thing. It's just automation. Uh, really cool to 101 00:05:47,089 --> 00:05:49,370 S1: go check that out, especially because it's book summaries, which 102 00:05:49,370 --> 00:05:54,680 S1: are useful complexities, allure and simplicity is power. Basically, this 103 00:05:54,680 --> 00:05:58,700 S1: article is arguing that simplicity is powerful, but complexity is 104 00:05:58,700 --> 00:06:02,240 S1: what actually sells and gets people excited. Uh, I'm I 105 00:06:02,240 --> 00:06:05,570 S1: don't agree with that for myself, but I think the 106 00:06:05,570 --> 00:06:09,080 S1: article made a pretty good case for it. 30 Useful concepts. 107 00:06:09,080 --> 00:06:15,679 S1: So I just found like 27,500 asteroids in old telescope photos. 108 00:06:15,680 --> 00:06:17,630 S1: And this is a good example of where you need 109 00:06:17,660 --> 00:06:22,010 S1: AI because there aren't enough people, professionals actually looking at 110 00:06:22,010 --> 00:06:25,100 S1: the sky with telescopes. There's just not enough people. There's 111 00:06:25,100 --> 00:06:27,679 S1: not enough eyes, there's not enough experts. So this is 112 00:06:27,680 --> 00:06:31,210 S1: a perfect. For. I actually just having access to tons 113 00:06:31,210 --> 00:06:34,060 S1: of different telescopes and then being able to just launch 114 00:06:34,060 --> 00:06:36,250 S1: an alert if it sees one that looks big and 115 00:06:36,250 --> 00:06:38,740 S1: it looks like it might be heading towards us, looks 116 00:06:38,740 --> 00:06:42,310 S1: like we might have found a potential for extraterrestrial life 117 00:06:42,310 --> 00:06:46,539 S1: by detecting a certain molecule. Looks like higher paid employees 118 00:06:46,540 --> 00:06:51,620 S1: are struggling, which makes sense to me because. If you 119 00:06:51,620 --> 00:06:54,140 S1: want to reduce how much you're paying in headcount, you 120 00:06:54,140 --> 00:06:57,680 S1: reduce the people first. That make the most. I've seen 121 00:06:57,680 --> 00:07:02,330 S1: this personally multiple times for people around me. So definitely true. 122 00:07:02,330 --> 00:07:06,469 S1: Got an interesting measurement of the economy, essentially how well 123 00:07:06,470 --> 00:07:10,730 S1: strippers are doing and how much they're getting tipped. Is 124 00:07:10,730 --> 00:07:15,630 S1: a good indicator or at least an interesting indicator. Number 125 00:07:15,630 --> 00:07:20,370 S1: one metric for longevity continues. Every study I've seen about this, 126 00:07:20,370 --> 00:07:24,090 S1: it's basically VO2 max. And this is yet another study 127 00:07:24,090 --> 00:07:28,140 S1: that's confirming that. Really cool essay. I didn't read it again, 128 00:07:28,140 --> 00:07:30,660 S1: but I've read this like ten times. In Praise of 129 00:07:30,660 --> 00:07:35,340 S1: Idleness by Bertrand Russell. And this one is really interesting. 130 00:07:35,340 --> 00:07:39,770 S1: We got this woman here who is complaining about. Oh 131 00:07:39,770 --> 00:07:44,900 S1: my God, they offered me. A help desk job. I 132 00:07:44,900 --> 00:07:47,870 S1: keep getting these offers for help desk jobs, and then 133 00:07:47,870 --> 00:07:51,320 S1: she proceeds to show off her two degrees. She's like, 134 00:07:51,320 --> 00:07:53,990 S1: look at this degree and I have a second degree. 135 00:07:54,020 --> 00:07:56,780 S1: Can't they see that I have these degrees? Why would 136 00:07:56,780 --> 00:07:59,330 S1: they be offering me a lowly help desk job? First 137 00:07:59,330 --> 00:08:03,350 S1: of all, help desk people are awesome. InfoSec Taylor Swift 138 00:08:03,590 --> 00:08:07,220 S1: was a help desk person. I know so many people 139 00:08:07,220 --> 00:08:09,740 S1: who started on the help desk. My buddy Jason Powell 140 00:08:09,740 --> 00:08:12,680 S1: started on the help desk and now he's doing amazing 141 00:08:12,680 --> 00:08:17,810 S1: things at Apple. And it's like, look, you can't. First 142 00:08:17,810 --> 00:08:20,840 S1: of all, you're demeaning a group of people who work 143 00:08:20,840 --> 00:08:23,900 S1: in a field. So that's a problem. Second of all, 144 00:08:23,900 --> 00:08:27,410 S1: it's not above you. And third of all, those degrees 145 00:08:27,410 --> 00:08:30,830 S1: don't actually mean anything for you, right? This is kind 146 00:08:30,830 --> 00:08:33,380 S1: of the problem, right? She's in a heart for a 147 00:08:33,380 --> 00:08:38,900 S1: hard time because. She believes that she's entitled to these things. 148 00:08:38,900 --> 00:08:41,059 S1: I do want to mention, though, it's not her fault. 149 00:08:41,059 --> 00:08:45,200 S1: It's the fault of the system for programming. And this 150 00:08:45,200 --> 00:08:47,750 S1: is like the parents, this is the teachers, this is 151 00:08:47,750 --> 00:08:50,990 S1: the whole school mechanism. And this tended to be more 152 00:08:50,990 --> 00:08:53,480 S1: true in the past. Right? You put the work in, 153 00:08:53,480 --> 00:08:55,969 S1: you pretty much get a job. It is just not 154 00:08:55,970 --> 00:08:59,000 S1: the case anymore. And it's definitely not the case now 155 00:08:59,000 --> 00:09:02,810 S1: with AI, it was already the case before. I mean, 2022. 156 00:09:02,840 --> 00:09:05,570 S1: It was not the case that a credential was good enough. 157 00:09:05,570 --> 00:09:08,929 S1: You already needed to be special in some sort of way. 158 00:09:08,929 --> 00:09:13,400 S1: So this was already a bad take, you know, back 159 00:09:13,400 --> 00:09:17,600 S1: in 2022. And now it's even worse, way worse with 160 00:09:17,600 --> 00:09:22,579 S1: AI because people like this who think that their homework 161 00:09:22,580 --> 00:09:25,820 S1: and their the fact that they finished these classes is 162 00:09:25,820 --> 00:09:29,120 S1: actually going to be valuable to that company by itself. 163 00:09:29,120 --> 00:09:30,950 S1: They're the ones who are going to get replaced by 164 00:09:30,950 --> 00:09:34,760 S1: AI or just not hired at all. And this is 165 00:09:34,760 --> 00:09:39,830 S1: another sort of thing around this. It's just like actually unrelated. But, um, 166 00:09:39,980 --> 00:09:42,830 S1: it's basically if you're in tech and you're constantly dunking 167 00:09:42,830 --> 00:09:46,070 S1: on AI, you should stop and you should think about this. 168 00:09:46,070 --> 00:09:49,790 S1: AI is about to add billions or trillions of new, 169 00:09:49,820 --> 00:09:55,490 S1: highly skilled and intelligent workers to the economy. And I mean, 170 00:09:55,490 --> 00:09:57,260 S1: this is what's going to happen, right? And I've got 171 00:09:57,260 --> 00:09:59,450 S1: a whole whole vibe here, but I'm not going to 172 00:09:59,450 --> 00:10:03,080 S1: read the whole thing. Basically, the takeaway is that something 173 00:10:03,080 --> 00:10:05,720 S1: about AI might be putting you off. Maybe it's like 174 00:10:05,720 --> 00:10:09,319 S1: the fanboys, maybe it's the fan. The fact that it 175 00:10:09,320 --> 00:10:12,530 S1: might sound to you like crypto or NFTs, you need 176 00:10:12,530 --> 00:10:15,559 S1: to push that aside and like, push away the fact 177 00:10:15,559 --> 00:10:18,650 S1: that you're you're turned off for some reason about this, 178 00:10:18,650 --> 00:10:22,250 S1: because if you were turned off about driving or turned 179 00:10:22,250 --> 00:10:25,970 S1: off about reading or writing it, just because somebody was 180 00:10:25,970 --> 00:10:30,320 S1: going crazy with CrossFit around reading and writing or reading 181 00:10:30,320 --> 00:10:33,829 S1: and writing, it's the best thing driving. It's the best thing. 182 00:10:33,830 --> 00:10:36,500 S1: And when you saw those people talk, it annoyed you 183 00:10:36,500 --> 00:10:39,110 S1: and you're like, you know what? Because of those people, 184 00:10:39,110 --> 00:10:41,840 S1: I'm not going to learn how to read, write or drive. 185 00:10:41,840 --> 00:10:44,920 S1: That's what you're doing with AI. That is what you're 186 00:10:44,920 --> 00:10:47,980 S1: doing with AI. And guess what? It's not hurting them. 187 00:10:47,980 --> 00:10:51,250 S1: It's going to hurt you. So that's why I'm saying here. 188 00:10:51,250 --> 00:10:53,950 S1: That's the point of this one. And the recommendation of 189 00:10:53,950 --> 00:10:57,670 S1: the week is my three minute video for how to 190 00:10:57,670 --> 00:11:00,460 S1: Build a meaningful life. It's pretty cool. It's like three 191 00:11:00,460 --> 00:11:04,630 S1: minutes 30s. And the aphorism of the week is one's 192 00:11:04,630 --> 00:11:07,330 S1: destination is never a place, but a new way of 193 00:11:07,330 --> 00:11:11,440 S1: seeing things. One's destination is never a place, but a 194 00:11:11,440 --> 00:11:16,540 S1: new way of seeing things. Henry Miller Unsupervised Learning is 195 00:11:16,540 --> 00:11:19,689 S1: produced and edited by Daniel Miller on a Neumann U87 196 00:11:19,690 --> 00:11:23,770 S1: AI microphone using Hindenburg. Intro and outro music is by 197 00:11:23,770 --> 00:11:26,980 S1: zombie with the why and to get the text and 198 00:11:26,980 --> 00:11:29,319 S1: links from this episode, sign up for the newsletter version 199 00:11:29,320 --> 00:11:35,010 S1: of the show at Daniel meisler.com/newsletter. We'll see you next time.