UL NO. 443: North Korean Co-workers, UBI Failure?, AI-Groupthink, GPS Spoofing…

powered by advanced AI. Over 7000 global companies like Atlassian, Flow Health and Quora use Vanta to manage risk and prove security in real time. Get $1,000 off Vanta when you go to Vanta comm slash unsupervised. That's vanta.com/supervised for $1,000 off. Welcome to Unsupervised Learning, a security, AI, and meaning focused podcast that looks at how best to thrive as humans in a post AI world. It combines original ideas, analysis, and mental models to bring not just the news, but

why it matters and how to respond. All right, welcome to unsupervised learning. This is Daniel Miessler. All right. So here's what's up this week gearing up for Las Vegas. Going to be there for 11 days. Assuming I don't get sick after like four of those. And if you see me, please, uh, please know that I'm a bit, uh, shy and awkward. A bit, uh, like 20 to 60% of the time. I mean, sometimes it's just not there at all, and sometimes it's pretty bad. So just, uh,

say hi. Anyway, I love, uh, seeing people, and, uh, it's a good training for me. Need to get better at this. Anyway, watch a number of videos last night, I believe, about people losing their jobs and starting, like, a YouTube channel and, like, just trying to branch out and maybe start a new career. And it was extremely sad. And I watched like two of these. And of course the feed showed me like ten more. And I ended up just watching these videos for like an hour and

a half, two hours. And it was extremely sad. I feel like people are seeing the ground shift underneath their feet and they're like, why does nobody want me, right? This one particular one was really sad because he was like, I thought I was a good product manager. And then after ten and a half months of interviewing and getting all the way through interviews and getting to the very end, and then not being given an offer or being given an offer and then being ghosted after ten and a

half months, I no longer believe in myself. I no longer believe I'm a good product manager, and it was just devastating to hear this. And he was being so like just open and honest and vulnerable. And it was just it was hard to watch and just felt very human.

And this is essentially why I'm doing what I'm doing now is to basically try to help people who are in that situation and try to hopefully explain what I see coming in terms of like, you shouldn't judge yourself based on what these companies think about your skills and what they think about your worth. Um, they do not determine your worth. And this is why we need something like human 3.0 is because human 3.0 is how to

measure yourself. Okay, you in your entirety. Full spectrum is how you should be thinking about yourself, not in terms of do you have these particular product management skills that this particular type of startup needs or this particular type of company needs? That is a very old world way of thinking about things. And I don't fault anyone for thinking that way because you're trying to get a job, you're trying to pay bills, right? So I understand that.

I'm just saying that society as a whole is stuck in the past in seeing the world in this way, and we need to move beyond it, especially because AI is going to accelerate this whole process where they're going to see more and more people as expendable in this way, the same way that this guy was fired in ten and a half months later, he still can't get a job. That's going to keep happening to more and more people, and we need a solution. So I would say be

kind to people. Everyone is hurting in some kind of way, and especially right now. And when you look at our politics, I see a lot of the current situation as a symptom of people being hurt and people being afraid. And I think that makes people more mean and less empathic than they otherwise would be. And I got this great quote here. Kindness is a language which the deaf can hear and the blind can see by Mark Twain. All right. So I got my best argument for why I will

have an extraordinary effect on economy and jobs. And if you have any friends who are still skeptical about AI, I recommend sending them this. It's called, uh. If we've been thinking about AI all wrong, created a video for substrate. It's, uh, one of my best performing videos, actually. Uh, and if you prefer videos to long articles and you still want to learn about substrates, recommend checking that out. Security Knowbe4 accidentally hired a North Korean state actor who tried to

install info stealing malware on their device. The craziest thing about this whole North Korea stuff and how they're like breaking into these companies, they're not really breaking in. They're applying they're applying to these companies and getting hired. And it really reminds me of this Key and Peele skit where the guy's like, yeah, so what we're going to

do is we're going to sneak in. I think I told this joke before, but he's like, what we're going to do is we're going to sneak in, we're going to get hired. And then when they give us work, we're going to do the work, and then they're going to deposit money in our bank accounts every two weeks. And his buddy looks at him and he goes, dude, you're talking about a fucking job. That's what a job is. And that was the end of the skit. And that's

literally what these guys are doing. They are applying to jobs, getting hired and doing the work. The trick is they just take their regular salaries, and I'm not sure what percentage is actually doing the work versus doing the work poorly versus doing the work, plus trying to hack them additionally. Not sure about that exactly. I'm not sure anyone is,

but here's what I do know. Uh, they're taking the money from the salaries and putting it back into the government to use for like the missile program or whatever other shenanigans North Korea is doing. So they're they're literally deploying it people into the world, into advanced economies to get actual jobs, to earn salaries, to put into the missile program. That is trippy to me. That is completely insane. All right. Next one GitHub's repository design flaw allows indefinite

access to data from deleted and private repositories. This is a piece from Truffle Security. A plane's GPS was jammed on a commercial transatlantic route for the first time, raising fears that thousands of other flights could be at risk of deliberate hacking. This was between Madrid and Toronto, thanks to teens for subscribing or not subscribing for sponsoring, and there's been a 400% increase in GPS spoofing incidents, so this one is related to the previous one affecting around

900 flights daily. And this is from ops Dot Group. And um, yeah, spike is causing major safety concerns, especially with systems like e.g. AWS becoming unreliable and there are a number of people looking at like, why is this happening? What are the implications? Yeah, I'm wondering what the other options are for Nav if there's like secondary or tertiary systems that we can use when a jam happens or blocks against the jamming, like a couple of different options.

The other thing is if there's six miles up or whatever and the jamming is happening, is that coming from inside the house? Like, is that someone with a jammer on the plane? Is that someone on the ground? I mean, these are kind of the only options. Someone on the ground is shooting it up at the plane, which seems pretty hard. Another plane coming along with it and shooting it. That seems very hard and also very expensive. Or they're in orbit, uh, and they would be moving very fast

if they were in orbit. So and they're shooting down I mean, I'm trying to list all the options other than aliens. And, uh, it seems like someone on the plane is more likely so out of those. But, uh, I don't know. I have no insider knowledge on this

one yet. Uh, Francis, high speed rail traffic got disrupted due to what officials are calling malicious acts just before the Olympic ceremony, thanks to nudge Security for sponsoring, Google has decided not to phase out third party cookies in Chrome and will instead offer users more control over how

the user's cookies are used. And I, for one, am absolutely shocked that this company that makes most of its money on advertising and is really bad at rolling out products canceled a thing that's really one difficult to do and two would negatively affect advertisers and therefore their main source of income. An evaluation of Amazon GuardDuty reveals limited coverage and high costs with significant latency in detecting attacks

like S3 ransomware. Amazon. When it does things like this, it's a little bit similar to Cloudflare, where they just kind of like do an initial version of the tool. It kind of works a little bit, and then if you adopt it, you just kind of hope it gets better over time. And basically what a number of these articles are basically saying is that, yeah, they don't necessarily always catch up. It depends on the product team, it

depends on luck. And sometimes if you lump your eggs in with that basket, like it just goes nowhere and other times it becomes a full fledged product. So other examples of this GuardDuty, the WAF product, there's a million different ones. Google's reCAPTCHA is showing its age and is harvesting user information with labor worth billions, while being almost

universally disliked and vulnerable to bots. The Senate unanimously passed the Defiance Act, letting victims of non-consensual intimate images created by AI sue their creators for damages and then get up to 150 K to 250 K if linked to assault, stalking or harassment. A US Commerce Department or the US Commerce Department says shipments of high performance processors from China and Hong Kong to Russia have dropped by 20%, but

Hong Kong is still the main hub. AI and tech whiz turned down $23 billion in acquisition from alphabet and is instead aiming for an IPO. I thought I did a deeper comment on this. I think I do a deeper comment on this later. Yeah. Oh, here it is. Yeah. My thoughts are basically that they knew they could get more money for this. That's number one. But two is they basically knew that if they got bought they would

get that 23 billion, which is a decent amount. And the problem is it would most likely end up in the Google graveyard. So who knows what percentage of that was. Maybe that description is online somewhere. AI is replacing jobs in the video game industry with major companies like Activision using generative AI tools for concept art, and they're saying, uh, estimated 10,500 people lost their jobs in 2023. I don't know if that's all from art. I imagine a lot

of it is from art. But, uh, yeah, I got a bunch of buddies in this industry and they're saying, uh, very similar things. Lots of different things in the creative workflow are vulnerable, I would say. A new study shows that while generative AI like ChatGPT makes individual stories more creative and engaging, it also makes them more similar to each other. And I've heard a lot of criticisms of AI and like, oh, it's going to do this bad thing or that bad thing, but this one sounds quite realistic.

So basically we need to engineer in mechanisms for exposure to alternative frames of reality. Different models, different ways of viewing reality to avoid people consolidating and following like one AI powered groupthink. Or maybe it's like 20 AI powered groupthink, but it's like controlling the narrative of how everyone's thinking about certain things. Because something was generated by AI, it went viral. Maybe it was pretty cool or whatever. So it goes viral, and then that just becomes a thing

that everyone's thinking. I've actually been really worried about this for myself. It's one of the main things I want to use ADR for. And I've been talking with Joseph Thacker about this as well. It's like, imagine that you have in your brain at any given moment a number of ideas, and I definitely have this. I have multiple ideas brewing all the time. So what percentage of those are like super new that I came up with myself?

What percentage of those are like mostly derivative? And how many of those are just seeds planted by some random add that I didn't even understand? And like this is completely opaque to me. Yet I think that these 19 ideas are like mine. And so I'm going out and sharing them with other people, and then the same phenomenon happens to them. And then the question is how many of those came from I? Then the question becomes how many came from a government or some corporation where they

injected these things into the world with AI. And now these memes are spreading all over, and pretty soon it's in all the articles, it's in all the news, it's in all the top podcasts. And essentially what you have is like a soft version, lower case version of propaganda, mind control and the steering of narratives. So I think it's going to be super essential for people who care about free thinking, want to be able to disconnect and sort of read books slowly, think slowly, come up with,

you know, old style thinking, old style critical thinking. That's going to be one thing. But the other thing is I want my Da to monitor the ideas that I'm talking about and that I'm writing about, monitor those things, and then also say, hey, guess what? So I saw this idea get released into the world by this whatever. This shady group over here. That was three and a half weeks ago. They then started broadcasting it on these

19 different networks. Those got picked up by these 347 other networks, and that became one of the top 14 top ideas over the last two weeks. And you are now over here writing about it. So I just want to let you know as my as your friend that you are being influenced by this campaign that was launched by this group three and a half weeks ago. That

is a service that we all need. That is a service I'm going to build for myself, because that's going to be part of this overall mechanization involving substrate pipelines, a whole bunch of other stuff that's basically monitoring not just the world, but also myself and seeing like how I'm being influenced. That's like one of the use cases. All right. Switzerland has passed a law requiring all public sector agencies to use open source software and open source,

any code that they develop. I love this, but, uh, open doesn't mean secure by itself. That's one thing. The other thing is you got to find a way to support the stuff, which you could just hire consultants or companies or whatever, but or trained internally, but definitely have to worry about support Govloop super cool platform for AI workflows doesn't quite work as far as I could tell. And if you're like a founder there or something, feel free to reach out and let me know. And, uh,

happy to correct that. But that's just my current impression. But I am extremely excited about it and I think they're doing great stuff. The guy is amazing. Reminds me of Yahoo pipes. Definitely want to check it out. Alphabet's putting another 5 billion into Waymo. It's basically looking like Waymo versus Tesla for self-driving taxis. But the approaches are

very different. Waymo basically needs to train for a very long time in a city before it works, because they kind of work out all the bugs just by driving all the streets for a very long time. And, and they sort of ramp that up. And Tesla is like typical Elon. He's like, no, I'm just going to make the thing. It's going to be awesome. It's going to work perfectly. I'm announcing it soon. But you have to

realize that he's not always right, right. He said full Self-Driving was going to be pretty easy and we would have it years ago. And now it's 2024 and it's just now getting good. So something to think about there. Between the two different approaches, Joe Procopio argues that tech companies are struggling to find good employees because they focus too much on creds and not enough on skills. I like it. I think this is exactly the shift that's happening.

He basically suggests that companies should prioritize practical experience and problem solving abilities over degrees and certifications. That's exactly what's going to happen with something like Astra, which is a fake thing. But this this article that I wrote that rates things in and gives scores, basically does a gauntlet of tests and then gives you like these comprehensive, really deep scores. That's essentially how that's going to be solved.

Apple just launched a beta version of Apple Maps for the web. Humans. Did I talk about ChatGPT? I thought I did I take that out? Where did that go? Oh very strange. Yeah, there's a new ChatGPT a search web ChatGPT that basically competes with Perplexity and Bing and it's it's out, but it's waitlisted. Um, so, uh, I don't know. I haven't checked to see if I have access to that one yet, but, uh, looking forward to

that one. Humans. Wall Street Journal explores why the US birth rate is declining, citing economic uncertainty, career priorities and lifestyle choices. The idea that UBI reduces the need to work isn't new, but recent studies show that it does not lead to better jobs or more education. I think the issue here is that certain people will spend free time and money to better themselves, and certain people won't,

and it's not clear why that is. It's not clear what the actual like factor is, but the way forward is basically trying to figure out what that is and not believing in fairy tales like giving away free money will make everyone ambitious. It does not. And this reminds me of a lesson that's very similar that I learned like 20 years of hiring people, which is exposing people to training and encouragement makes the best people stand out, but it does not turn everyone into the best people.

And I've known this because I tried because I tried for like ten years with cohort after cohort where I would just be like, oh, well, my training wasn't good enough. That's why they didn't get it, or they were distracted or they were busy or whatever. So I just keep hitting them and keep dumping hours and hours and hours and more training and more training. Send them to more classes and you come back to them and you're like, hey, what have you done? They're like, I don't know. I

didn't do anything. What should I do? And you're like, uh, no, no. So you've you've got to figure out what you want to do. You've got to I've given you all the training, I've given you all the stuff and all the books, and I gave you all the demos. I told you how to set up the demos. I showed you how to set up the tech, and, um, have you done any of that? And they're like, no, what should I do first? I don't I don't know how to do any of that. Can you show me? And you're like, well,

I showed you four times. You want me to show you a fifth time? Okay, sure. I'll show you a fifth time. Show them a fifth time. Come back two weeks later. Hey, how is it going? Going with what? With the stuff we talked about last time. And they're like, yeah, it never got around to that. Like, I couldn't find this one driver. Like, there was a driver down low, but the driver download didn't work. It's like, okay, well did you research that? No. How would I do that?

Can you help me that That ends up being like 50 to 80% of everyone. Here's the alternative. Here's the alternate version of a star. Okay? You're like, hey, um, hey, everyone. So check this out. This is a class about. And you see a hand go up. Hey, look, I read the whole document. Um, I looked forward in the class. I've got this thing running, but there's an error. Um, so I fixed it and I stood up a different version,

and it's running. Can I show you? And you're like, okay, so you basically read everything, jumped ahead in the class, built something. It didn't work, but you fixed it, and now you're running a different version and you want to show it to me now. But I just started my first sentence of the class and they're like, oh yeah, sorry. Yeah. Please continue. Sorry. I was just excited. The difference between

these two things are so insane. Okay? And what I've learned after all these years is that if you're trying to build a star team and you're trying to build like a group of the best people, what you want to do is throw help and throw training and throw encouragement to a massive group of people and not judge, not prejudge, not do anything you have no idea. You have no idea. It's it's not people who wear yellow shirts. It's not people from Idaho. It's not people from Mississippi.

You have no idea what the criteria are like. Background, ethnic group, you education, background. Are they educated? Do they have a PhD? Do they have two master's degrees? Are they still in high school? You don't know. And that's the smartest thing you could possibly realize is that you

have no idea. You have no idea who's going to just be kind of like, complete, completely inert, and just unable to learn and unwilling to learn and unable to absorb any amount of effort that you give, versus some random person who doesn't think much of themselves and are unassuming and quiet, and they just absolute sponge. Absolute murder. Like the the craziest, best IT person you've ever found.

And suddenly they just can't shut up about everything. I was like, oh, and I did this and I read everything, um, all those books that you recommended. And then I went and downloaded this and I oh, but I changed it, and I want to show you this thing, and you're just like, Holy crap. They were sitting right next to the other person who I gave the same exact training. This is a lesson, incredibly powerful lesson. And to bring it back to this article, I believe the same thing

is true with UBI. The same thing is true with social programs. The same thing is true with all these different things. So the whole trick and we're going to get a little broad here. The whole trick is to make sure that no one's at a disadvantage, okay? Because trauma, because they're hungry, because they can't afford coverage. Right. Because there are stars all throughout this entire population, this these

8 billion people that we have. There are stars who you will never know if they're a star because they can't make it to your class because they're working seven jobs and they got an ant and a dog and a and a parent and a girlfriend or a boyfriend who is sick. And so they're over here juggling and they're another Einstein, they're another von Neumann. And you will never know because they got screwed. They got screwed by luck.

So the goal and now we're going really broad. The goal is to build a society in which nobody has screwed in that way. And when they do get messed over in that way, we help them. We help them get back into this, remove the obstacles so that everyone has the opportunity to be one of these stars to stand out. But we should not think that everyone who gets exposed to the opportunities is necessarily going to be

one of those stand out people. So if we see variation, which is probably going to look a lot like a bell curve, right, you're going to have most people in the middle, um, when you see that variation, you should be like, yep, that's exactly what we expect to see. Most people in the middle, some people who don't want to do anything despite any training, and some people who could just hear the first word and write a book. Right.

That's just expected. Now, I would say that there's some nuance here in the sense that, like, okay, if we can identify that grit and self-discipline and I don't know, something that we could try to put into culture, something we could try to put into education systems, something that we could try to put into training, something we could try to put into parental education if we could find those tokens and just like, make sure they're part of

the training and that helps lift everyone or some number of people, that'll be fantastic. But it has to start with realizing there's going to be a distribution and that's going to be based on meritocracy, which is good. However, the whole purpose of society and liberal society or liberal liberal approach to building a society is making sure that everyone can enter into the meritocracy funnel without disadvantage. That's

that's the goal. That's the trick. And I'm going to actually try to instantiate this using substrate, by the way. All right. That was a oh goodness. That was a massive diversion. Uh, southwest is ditching its open seating policy after 50 years to boost profits and meet customer preferences. 80% of their flyers preferred assigned seats 80% is a lot. The Senate's version of 2025, NDAA, does not include the

countering CCP Drones Act. Happy and sad about this I love DJI drones, DJI drones, but I feel like we need to go without them because they're so good that there aren't many alternatives. And, uh, Chinese companies have the ability to be just taken over by the CCP at any given moment, if it's advantageous to them. So I don't like that. Same reason I don't like, uh, TikTok.

As you might expect from an ex-military security person, nearly 40% of Americans are stressed out about making ends meet, 40%, up from 28% in 2021. And this is a very similar number to during the Great Recession, which I believe they're talking about 2028, I'm sorry, 2008. US economic growth hits 2.8% weight loss. Drugs like Ozempic, mounjaro and Wegovy are causing people to spend less on groceries. I heard it's also making planes lighter and use less fuel. So

Wegovy is a climate change product. They should try to get some funding for that. Subsidies or something. A new antibiotic. Oh my goodness. A new antibiotic from the University of Illinois, Chicago disrupts two different cellular targets, making it 100 million times harder for bacteria to evolve resistance. One dose of a new nasal spray treatment clears toxic tau proteins from brain cells. And one thing to remember about these I'm going to keep sharing these studies like I am. That's

the thing I'm going to do. I'm going to have more and more scrutiny because part of the pipelines, uh, workflow and substrate and all of that, I'm going to have these things being tested and going through pipelines. It's going to be awesome. But you still have to realize the bigger the finding, the more you should wait for supporting studies. And in my mind, the stuff is not actually truly real until the drug is available to normal people,

which means it's already been rigorously tested. And of course it will be constantly tested afterwards as well in in the larger population. So that to me is like the real stuff. But I still like watching this. I always thought that this tau protein stuff was ripe for some sort of chemical solution, some sort of, you know, uh,

medicinal solution. Uh, of course, that was just intuition. I have no expertise there, but it just seems like something that people can grab on to once they realize the mechanism. Liberals and conservatives are both prone to conspiracy theories, they just prefer different ones. Hendrik Carlson talks about how generating interesting ideas is like building a muscle. He says the more you write and think deeply, the better you get at coming up with new and meaningful thoughts. 100% agree ideas.

Zuckerberg is arguing that China is going to steal weights anyway, so there's no way to stop that, so you might as well develop advanced AI as open source. Okay. Not really. Okay. But I think I wrote about that somewhere, so I'm not going to go into that. Search GPT oh that's where I put it yeah. Search GPT in discovery here I use obsidian Jason Pepper Hepler talks about how he

uses obsidian for notes. In the beginning was the Command Line, Neal Stephenson's classic essay on operating systems, my obsidian notetaking workflow. It's on my mind. So that's why I got multiple things in here on it bash one liners, data chain, unstructured data management for AI projects. Llama agent stack. This one's cool. I JSON has a cool, um, video on this.

You should check out on YouTube. Open world exploration in Minecraft Claude engineer la Cara low latency AI application firewall GPT four Captcha Bypass Flow Analyzer a tool for understanding OAuth two grant flows with support for Oidc and Jwts Jots. I don't know how you're supposed to pronounce that JWT. I bet you about 5050 people do that jot versus JWT bash, simple curses, simple bash library to create terminal interfaces and the recommendation of the week. I'm going to

try to do something for the next several months. Kind of do this already, but I'm going to try to do it even more going into this election. When someone labels me as super liberal, I'm going to say something nice about their conservative views. I'm going to humanize them. When somebody labels me as super conservative, I'm going to say something nice about their liberal views. I'm going to humanize them. Try this and see if it opens up

the conversation at all. I learned how to do this from Jonathan Hite's book, The Righteous Mind, by the way, and this was probably 6 or 7 years ago now. This book brought me to the center politically. It brought me to the center, specifically what it did because I was already pretty far left. I've always been left, um, always voted Democrat. I've only been a Democrat, I guess, uh, technically and legally my whole life. Uh, although I wouldn't say I'm a Democrat, I would say I've always been

liberal my whole life, but technically, yes. Democrat. Um, this book. Well, let me just continue on that. Born and raised in the San Francisco Bay area, which grew up with super liberal hippie parents, and I, as a result, have always seen conservatives, or I was trained to see conservatives as, uh, holding progress back in, in a negative way. So not just conservative, but conservative in a bad way. In other words,

conservative was bad and basically just viewed them poorly. What this book did was show me this is the, I think the main premise from the book, which I'm I'm happy for him that I remember this so clearly. I think the main premise is that both sides value things very closely, very dearly, um, in a sacred way. But they are different things. So it's not that one is going for good things and one is going for bad things. It's that both are going for good things, but they're

different good things. So the whole trick to being in the centre and the whole trick to moving progress forward is realizing that people are just different. Okay? And there there are extremes to this that make it not true. But in general, you should look for this common ground and realize that there being a good person, um, in their way and you're being a good person in your way,

and you're going to try to find, uh, Harmony there. Now, in the case of extreme sides of left and right, you can argue that, yeah, their good is actually bad for the whole world. So like, you don't want to support that. So understand there are situations like that. But in general we would all be better off by reading this book and realizing this, this nuance, which might help bring you towards the center and produce better conversation. And the aphorism of the week, the highest form of knowledge

is empathy. The highest form of knowledge is empathy. Bill Bullard. Unsupervised learning is produced and edited by Daniel Miessler on a Neumann U87 AI microphone using Hindenburg. Intro and outro music is by Zomby with the Y, and to get the text and links from this episode, sign up for the newsletter version of the show at Daniel miessler.com/newsletter. We'll see you next time.