All right. Welcome to episode 486. This is Daniel Missler. All right. A lot of stuff to share this week. Let's see here. Um, okay, so I had a debate with Marcus Hutchins about I, I don't know if you've seen him talking about AI online, but he basically, uh, him and a number of my friends, uh, Christopher Hoff being one of them, they basically take opportunity whenever they can to to basically point out how stupid it is and how it makes tons of mistakes and how it's basically all folly and, uh, how everyone's overestimating it. And the whole the whole thing is going to come crashing down and a bunch of scams, all for the purpose of making money. And it's all just hype and marketing. So that's kind of their position. Uh, Marcus is a particularly good at making these points in a very humorous way. And of course, he has a huge following because he is Marcus Hutchins, probably the most famous malware researcher that we've ever known. And, uh, because I have the opposite view, I wanted to have this debate. So we had the debate. It's on YouTube now. I highly suggest you go check it out. And of course, the link is in the newsletter. The conversation went extremely well. It was very cordial. It was very, um, substantive. I think we both made points that the other one received and appreciated. So I think it was overall just great. Now, a lot of the feedback has been that we weren't actually disagreeing and that, uh, both things could be true. And this makes me think we went a little bit too hard on being nice. And, uh, maybe for part two, which we're already talking about, talking about doing maybe for part two, we need to sort of tighten it up, um, sharpen the pencils a little bit and just be like, hey, look, you know, I think you're wrong about this particular point. Let's let's debate that. And, um, I think we got there. I think we did that. I mean, it was over a two hour conversation, so hopefully you could take that away in the conversation, but I think in part two I at least myself, are going to try to be a little bit tighter with my arguments and a little bit more direct. And, um, I don't want to say confrontational, but confrontational in the most cordial way possible to basically try to pin down these arguments and say, no, you know, this is where I actually do disagree with you. I think you're mistaken about that. And I think that will just be more useful to people who are actually trying to make up their minds. And most importantly, I think it'd be more useful to Marcus and his points and to myself and my points of like, if he thinks I'm wrong about this and I think he's wrong about this, we're ultimately trying to provide a service to our audiences and to audiences we haven't talked to yet. And, you know, future audiences. We're trying to make that point. We're trying to, you know, um, not just have a discussion, but but also try to show that we're right about this discussion. And it doesn't mean we're right about everything. It doesn't mean we've got it all figured out. We don't. It's the future. You can't predict it. But I think you see my point. So that's what I'm going to try to do in part two. And after that, I might do a specific sort of argument breakdowns or something even more tactical than that. But in the meantime, I recommend you go check out the episode. It's quite good. All right. Cloud code definitely my most holy crap. Are you serious? I tool that I've ever used. And again, I talked about this. I've been talking about this for a couple of weeks now. I'm telling you, like Klein and Cursor, those all still feel powerful. And I still use cursor sometimes. But here's the distinction I. I feel like cursor is a tool, whereas Klein. Well, and Klein is very similar, although I'm not using Klein as much. I'm using cursor now, and I would say that those are tools as opposed to cloud code is more like a coworker. It's more like another engineer that I'm working with, and I'm just kind of like sending work over there, like like I'm a senior engineer and they're a junior engineer, and I'm sending them work, and they're going and doing tests for me, and they're coming back and giving me a status report saying, here's what I got done, right. It's really, really that good. And I'm just blown away by it. And I can't even imagine, like, what gets better from here. So I talked about this metric metric. I think I mentioned it in this week's show or the one previous, but it's a metric for how long term of tasks I can actually accomplish. I think it's a really, really cool metric metric. I thought they might have been mispronouncing or misstating miter miter, but no, it's metric. And it basically talks about, you know, the fact that I struggles with long context problems with large scope problems where it just gets confused and just can't remember its place, can't remember all the different subtasks, and it just kind of spirals into like chaos. And that's a huge problem for AI, and it's kind of the big bottleneck right now, along with context, which is very much related to this, and along with working memory, which is very much related to this. So metric, this metric, I believe it's a pretty good capture of this overall collection of problems that I currently has, and how good of a job we're doing at addressing those. So all that to say, I think Claude, is Claude code is the best right now at actually dealing with this. So something to keep an eye on. Highly recommend this conversation with Arthur. Arthur Kroeber and Dwarkesh Patel on China. So, Arthur Kroeber or Kroeber? Sorry, Arthur Kroeber. That's a little bit difficult to say. So he's a China expert. The the China expert that everyone's talking about right now, or at least in my circles, like there's a lot of video of, um, Peter Zeihan and he's just talking about how China is so screwed and they, you know, their population collapse and, you know, they just they're in really bad shape and they're going to be done any year now. And I believe he's so popular because everyone wants to hear that. I definitely wanted to hear that. I was definitely happy to hear it. And what it did was kind of expose my bias because it's confirmation bias. It's also good news. And those are both very infectious. They kind of burrow into the brain. And I'm a little bothered that it worked so well on me for so long for like a couple of years where I'm just like, oh, this guy's super bright and China is actually super screwed. Well, it turns out if you ask a lot of people and if you sort of dig around, a lot of China experts and a lot of economists don't think Zion is all that smart. They don't think he's all that educated and knowledgeable about China or China's economy. And some of that could be hate. And this is why I was a little bit confused. And maybe I am still a little bit confused. I'm open to the fact that Zion actually still does know a lot of stuff. And I think that is the case. The problem is, I've seen him talk about cyber, and he basically said that, um, the Facebook app is key, logging every single keystroke that you do at all times, even if you don't have it open and it's uploading all of those keystrokes. Everything you're doing on your computer, up to Facebook at all times. And the problem is, he said that with the same exact conviction that he's talking about with China. And I'm like, what? Wait a minute. So I have to go and review all the all the dumb shit that he said about China, which I have been absorbing. And I'm like, this guy, is this lost about this? And it's not just some esoteric, like one philosophical view versus another. He's talking about a very tangible thing. Like, to me, this is like lying about a stat or something. This is like he's claiming that this is technically true when I know it not to be true. And just just as a simple way of like giving an argument why I know this not to be true or believe very strongly, at least if the Facebook app was up stealing and uploading the keystrokes of all its users all the time. Do you not think that this would be a huge story? Do you not think that all the security researchers out there in the world would not have noticed this? Right? Do you think Facebook would have meta would have okayed this and actually thought it was okay and it would never be caught? Like that is just ridiculous. So he's claiming to have this knowledge that no one else has and he's spreading it, you know, and I'm just like, wow. Right. This is where you have to really question things. So anyway, this other person who came on the Dwarkesh podcast, this guy Arthur Kroeber, is a lot more serious. You can just instantly tell, like, he'll be like, well, I don't know about that. I wouldn't go that far. Or, you know, like he's not talking like an influencer where it's just like, yeah, which is what zeihan it talks like, you know, if you listen to it, all his statements are absolutes. Um, that's not completely true, right? He, um, he sometimes says, oh, we don't know for sure or whatever, but a lot of his statements are absolutes. Um, and it's just like, first of all, we don't know that much about China. Like, the data is sometimes hard to know, right? Uh, especially because we're going off of their claims or we're going off of inference. We're trying to figure these things out. Um, bottom line is, he's just way too sure and way too confident about opinions that I'm not sure that that he should have at that level of confidence. And, um, it's great to hear someone on the Dwarkesh podcast. Notice he didn't have Peter Zeihan on. He went and found this guy who's a well respected economist, uh, and who he covered so much ground. I mean, it was just a fantastic conversation. And, uh, I would say the short version is I came away thinking China is in much better shape than Zeihan believes. And then a lot of people believe, and it's more in line with what my concerns were of where China might be. And essentially, Zeihan was giving me some confidence that I was wrong, that China wasn't actually in that good a position, but it seemed obvious to me from like other economic markers that they actually are thriving and they actually are, you know, in a position to kind of dominate the US, uh, in, in short order. So I would say, go listen to this. If you care about China, go, go listen to this. If you care about national security, economic security, the future of the US, the future of China, this conversation is spectacular. And it's partially because of how good the guest was, but it's largely because Dwarkesh is so good at interviewing. He is just very young, very smart, very curious. That's the biggest thing is like incandescent curiosity. That's the thing with Dwarkesh. He's just like, he can't wait to jump in with his next question. He's moving the conversation along, like his conversations with Tyler Cowen are some of my favorite podcasts to listen to. Anyway, it's a fantastic episode. Okay. Very fascinated by this. Uh, what do you notice, Psyop that I keep seeing, especially on X, but it's on lots of different social media, I would say. And what it basically does is it shows you like the picture of, like a nine over 11, uh, building falling. Right. One of the twin towers falling. And it's like, watch it again in slow motion. What did you notice? Does this look normal to you? And what it's doing is like seeding this this concept of like, question everything. You've you've been lied to and it's like really, really pernicious. It's nasty. Um, the, the other cool thing that it does from a marketing standpoint, it requires you to watch the video all the way through in multiple times. And I think a lot of people do. That's why a lot of these things have like millions upon millions of views. And maybe that's just the propaganda people pushing it, but it implicitly asks you to use your own special analysis skills to figure out what's wrong with it. And gullible people love to believe that they're experts in lots of different stuff. So they're like, oh, you know, I'll use my special skills to figure out what's wrong with the building. He's like, oh, it shouldn't have fallen that fast. And like when there's a comment, it's like, did it fall faster than it should have? They will suddenly pop into their mind. Yeah. It did fall faster. That's how that's not how fast buildings are supposed to fall. Somehow they're now an expert in building falling like velocities. Right? And this is just like conspiracy thinking. This is what a lot of people have. Um, and I would say everyone has it. It's a question is like, how much do you have it in check? Right? And then they paint the perfect picture in your brain without actually saying anything. Sometimes they say it in the comment, sometimes they hint at it, but ultimately. Right. The brain watching tries to like, square the circle, coming up with an explanation when there isn't one. Or there might not actually be anything wrong with the video, but because they told you there was something wrong with it, you still are these gullible types who watch it leave feeling like, ah, you know, they have been lying to us. That video was crazy. I can't believe I saw that video. It's like, no, it was just a regular video, just regular iPhone took a picture and sent it up there. It's like it just opens the opportunity for gullibility and conspiracy thinking like I've never seen before. It's a brilliant tactic. So basically, the conspiracy brain goes on special forces mission to solve the puzzle, and then it's followed up with like over the course of a day or a week or a month or whatever, like hundreds more videos hitting their feet. 1 or 2 of them will give you like an explanation for this thing. It's like, oh, that's because they're doing this. That's because of, you know, it's all a conspiracy. That's because of this or whatever. And in maybe like one out of the ten or something, they'll give you a reason. But in the other nine it doesn't give you a reason. It just asks you to think about it and what the brain does, what, in my opinion, these people's brains. And I'm putting that in quotes. What they're doing is they're filling in the blank for the other nine videos, right? They're like, I solved it. I know the answer. Um, Um, I figured this out. This is a conclusion that I've come to myself not realizing it was either fed to them or they jumped to it, and they're now sticking with it, right? And it's just like, this is so nasty. And notice that the videos get millions of views. Like, how are they getting so many views? And so basically, like here are the narratives that I'm seeing. And I'm starting to bookmark these things. I don't like them because I don't want to like, um, encourage them. Uh, but I bookmark them. So I could actually, uh, put together some research on these accounts. But, uh, it's things like whether it's changing, food is changing. The world is basically different now in some inexplicable way, which you should try to figure out. Uh, women aren't women anymore. Men aren't men anymore. They used to be us. Used to be on top. Now we're just a bunch of pussies or whatever. Um, black people are nasty and violent. Nobody likes them for good reason. Anything that's wrong with you, it's from vaccines. US is the most racist country ever with constant crimes against non-whites. So the stuff is designed to anger you. It's not pointed towards or against any particular group. What it's doing is finding all the trigger issues and just blasting them right, with millions of views, which that might be paid promotions, that might be the fact that it's just viral and people are resharing or whatever. Um, liberals and black people and gay people and LGBTQ people have basically ruined the country. Russia is way nicer than the US. It's full of pretty white girls and no black people. You've been sold a lie that Russia is bad, but people love it there. They actually love it there. And most importantly, you are being lied to and you should be angry about it. And this is the message, right? It's like everywhere, but especially on X. And there are like hundreds of these accounts. And if you click on any of the accounts and you just scroll down, it's nonstop. Just these types of posts and they are formulated with like the same sort of quality and the same sort of narrative hooks. It is completely obvious to me that these are all part of campaigns. So the question is, who's paying the money for these campaigns? Who is creating this content? And my answer is it's absolutely propaganda. It is enemies of the United States who are doing this. Probably mostly Russia, but there could be others. It might even be like internal groups who are just trying to stoke anger for particular internal candidates. I have no idea. I haven't done the research on exactly who these groups are, but I think we know from the past that it's likely to involve Russia to a decent amount, but they're not the only people doing propaganda. And the bigger point is that AI is also going to make this so much easier. Like what happens when you. I can write this stuff. I have no idea if I actually did write a bunch of this stuff. Maybe that's why I'm seeing so much of it now, because it's more scalable. But I would love for someone to go and take a look at all these different accounts and actually figure out who they are. See if there's any, you know, Osint like, uh, clues in there that we can, we can use to track them down. I want to know which countries are doing this. I want to know how much they're spending. I want to know how they're getting so many views. I think it's fascinating. just fascinating to see how propaganda works. And um, one other caveat I want to mention here is that I'm saying this as an ex-military person who was an Intel like a little bit. I basically spent a long time studying influence operations in the Army. When I was assigned to S-2 for my battalion in the 101st airborne. And does that make me like an absolute expert in influence operations or propaganda? Not really. I've read a whole bunch of books on it. I studied it when I was in the military. Um, I was sort of assigned to doing counterterrorism type influence stuff a little bit when I was actually in the military, but I'm not like a trained Intel person from the military. I was, uh, airborne infantry, so I was assigned to it because I seemed to have a talent for it. But it wasn't like I'm not formally trained in it or whatever. So it just just keep in mind, I read a whole bunch of books. Doesn't mean I'm an expert. Um, what I'm trying to do is to just get you to see my perspective and how I'm seeing this very clearly as being part of campaigns. And everyone who I know who actually is trained in this stuff. Um, as much or more than I am. They also agree other people trained in this way also agree that this is propaganda being sent against the United States. From who? Who knows? Right. You have to do that research. Uh, but but I would say just watch for it. Um. Oh, and I've got some questions here. Have you noticed any of your friends feeling these feelings of, like, America is screwed? Oh, I hate that group. That group ruined everything. Um, on X particularly, but also just in, uh, social media in general. Have you noticed yourself feeling more angry towards other groups or towards, um, just like the past used to be better and feeling like, yeah, Russia had it figured out because Russia seemed super nice or whatever. And are you consuming a whole bunch of social media? All this to basically ask, are you feeling yourself being influenced in this way? right? Because that's what I have to ask myself all the time, because I'm constantly trying to clean myself of bias. Not always effective. And I end up catching it like I've talked about multiple times in the newsletter and on the podcast. I'm constantly checking myself having bias. It's the other reason I do the Telos thing is to constantly be able to ask I and say, what mistakes am I making? What biases do I have that I seem to be, uh, you know, ignoring. So anyway, watch for it. Think about it. Uh, something to keep in mind. Uh, traditional advertising is in serious trouble. This is something, uh, just incredible. There's some ads of, like, um, the ones I particularly like, these I ads, uh, from Vo3 from Google. They're basically Bigfoot ads. Uh, Yeti, Bigfoot and Yeti ads. And they're as funny as, like, Super Bowl ads, which they pay millions of dollars for. And a number of groups have made these things, and it does actually take some talent to make it because you have to write the script, you have to write the prompt. So I think there is comedic human talent going into these. The question is how much and how long will that last? But I find it really interesting that we're seeing these things have like hundreds of millions of views. Okay, these Bigfoot little videos which are sometimes are selling things and sometimes they're just comedy skits or whatever, hundreds of millions of views. And I think a lot of advertising is going to move in this direction. There's one selling like a night vision goggle, and I'm like, I think I kind of want it. Like it didn't say anything about the night vision goggle. It was just funny. And I'm like, where do I click? Where do I buy it? Also, I need a night vision goggle. Um, I don't actually. So hopefully I won't go and click that and buy something unnecessary. All right. Cybersecurity A top hacker on hacker one is a fully automated AI agent called called XPO. And it's actually this is a slight correction. Got a little errata here. live. It's actually the top hacker in the US, so there are higher scores globally, but the highest US score is from a fully automated AI. And that's beating out all the top hackers in the US. That is ridiculous. Absolutely ridiculous. And this should not be possible, given what Marcus believes about AI, for example, or other people who don't believe that AI is intelligent, to be able to go and research and find and illustrate and actually submit a report that gets you results on one of these platforms, you have to you have to show that it's actually a real vulnerability, and it has to not be a dupe. You have to be first, and it has to be a significant issue. Right? We already have a high standard for what gets accepted. You know, Bugcrowd has this, Hackerone has this. So this is hacker one, accepting these actual vulnerabilities from actual AI. And it's number one in the entire United States. So anyone who thinks you can't automate this stuff, it's not possible or it's going to take years, it's going to take three years or five years or ten years. Nope. It already happened. It already happened. Number one hacker in the US. Uh, for for bug bounties, for mostly web vulnerabilities. Keep that in mind. Anthropic says multiple models resort to blackmail to avoid shutdown. Russian hackers beat Gmail toufar with app specific password social engineering. So this was, uh, intelligence group. Was this. Yeah. Russian group. Apt29 trick targets into generating Gmail app specific passwords by sending fake State Department PDFs, giving them instructions. And they filled it out. Stargazer's trick Minecraft players into installing malware through fake mods. 500 fake GitHub repositories to distribute malware disguised as Minecraft mods. 16 billion credentials leak. You've probably seen a million stories about it. It's complete garbage. It's just concatenation. Combinations of previous breaches. I don't know how so many big names like security outfits, media groups like totally just I guess it was so exciting they couldn't resist the the headline. Anyway. Quite sad. It's a nothing burger. Minnesota shooter allegedly used data brokers to find victim addresses. Uh, good time for a reminder that data brokers are way nastier, at least in my opinion, than the dark web, because they have more data, they're way more organized, and they're actually legal. They're collecting and selling this data legally, and they have way more data on all of us, and it's way higher quality and it's way better organized, like the dark web. And criminals have nothing on this, and it's completely legitimate. But we don't talk about it because it's legal. That's ridiculous. Aflac gets hit by a scattered spider in ongoing insurance company attack wave. So, uh, yeah, this group scatters spider. They kind of focus on areas, and they moved off their previous one. And they've been going after a lot of insurance companies. Apple and Google keep selling Chinese VPNs that could spy on you. So someone found a whole bunch in the Apple App Store and in Google that were Chinese VPNs and associated with, uh, not good groups, I believe. Chinese government. Um, I'm not sure about that. It was associated with something clearly malicious, though. Um, although I would just say if it's a Chinese company, just kind of assume that, uh, national security OpenAI gets $200 million defense contract to build AI tools. Yeah, Pentagon gave them $200 million to develop AI for national security and war fighting. I've said a lot on this before, so I'm not going to say it again right now. I'll talk about it again later. Uh, this is good and potentially bad. China's military is using AI for intelligence operations. Recording future recorded future found that China's People's Liberation Army has moved beyond just talking about AI for intelligence is actually procuring and deploying generative AI tools for military intelligence work. They're using it for like, uh, processing satellite imagery, generating intelligence reports that just basically like augmenting their military using these tools, including OpenAI. Iran's using hacked security cameras to guide missile strikes on Israel. Russia's deadliest Kiev attack this year kills 15 as peace talks have stalled. Kind of a kind of a lull and an advantage for Russia. I think, uh, with the stuff happening in Iran. Former Army sergeant shows how not to be a spy. So he used Google to search a, quote, countries that don't extradite to us and, uh, emailed China directly, uh, to share this information. Uh, he's sharing this, uh, classified information, uh, to, to send it to China, but he's using his own personal email accounts. Yeah. Uh. All right. I don't know why that's as funny as it seems to be. Taiwan gets Ukraine tested drone software to counter China. ex-CIA analyst gets 37 months for leaking Israel's Iran attack plans. A CF William Rodman photographed top secret documents about Israel's plans to attack Iran, edited them to hide their source, and then shared them with people before the documents went went viral on telegram. Deep tech allegedly using shell companies to support the Chinese military. US pressures Vietnam to remove Chinese components from tech manufacturing. I meter okay, that's the one I was talking about. Better metric for AI and agents. It's basically yeah. Long term tasks I think is really smart. Uh, Andrej Karpathy says prompts are software 3.0. So we did a major talk two weeks ago and basically said that software 3.0 is prompting, Prompting to an eye to have it build the application for you. And he calls the software 3.0. I love his really high level, you know, thought leading type of talks, and this was definitely one of them and absolutely loved it. I love his other talks too, but I love talks that are about, you know, thinking about things differently. He also said, we're in the decade of agents and not the year of agents, which I think is really smart. Sam Altman says GPT five is coming in the summer. Google uses YouTube videos to train Vo three without creator consent. Google also released a stable version of Gemini 2.5, which includes new models with different sizes and capabilities. This one is not in here now, but they also just released a command line tool, which I haven't messed with. I'm going to mess with later today as soon as I'm done recording actually. And um, yeah, it looks like maybe a competitor to Cloud Code, although it looks like it might be more of a general agent. So we're starting to move towards like the Da stuff that I've been talking about forever. Wicks bought an AI that lets anyone build software by chatting, so they bought a company that is an AI tool for building software. So this is starting to become a thing. Oh, the other thing I want to mention right now is there is a company called This is Major. There's a company called Cluley, and it is doing very much like what I wrote about in 2016, where it's like a digital assistant, but it's not just taking tasks from you, it's actually being more proactive. So it's actually recording your whole screen. It's doing it's watching what you're doing. And you could just actively ask it things. It could go and do things on your behalf like more proactively. It is kind of like you remember the Microsoft recall thing. It's kind of like that. There's also Google has moved towards this. Anthropic has moved towards this with like computer use. Everything is moving all in the same direction, right? It's all moving in the same direction of your Da is the thing that understands everything about you. I just read my book right before I came on this. I was talking to Joseph Thacker, one of the best hackers in the world. And, um, yeah, I was talking about I'm like, dude, you got to finish reading this book. So I went and read it real quick and it takes like 20 minutes to read the whole thing. It's 17,000 words, but it goes really fast. It's it's written pretty simply and it's online for free, right? You don't have to go and buy anything because that's ridiculous. But, um, anyway, it's all pushing towards what I was talking about in that thing, and I've got multiple examples in there. It's basically it knows everything about you and it's constantly advocating for you on your behalf. It's constantly trying to adjust the world around you using the API ification of everything. Universal demonization. Right? Everything gets an API. This is what's in the book. All companies get an API. I should stop saying book. I should start saying it's in the piece or it's in the content or whatever. Um, companies get an API, people get APIs, objects get APIs, businesses get APIs. Your Da speaks API, right? That was kind of the core thing for the for this, for this concept, for this book in 2016. Right. It's like your Da speaks the language of these other things. Does that sound like anything to you? Does that sound like agent to agent? Does that sound like MCP? MCP is basically an implementation of universal demonization. That's what it is. And I said right there in the content, it's like, look, I have no idea how this is going to work, but it's probably going to be HTTP based because the predictions I'm making there are about direction. It's stochastic. Right? It's not about like who's going to win whatever. You can't predict tech like that anyway. Cluley is heading directly in this direction. Um, I'm wearing a pendant right now, which is recording everything that would be input into my Da. My da's name is Chi, by the way. Um, haven't fully spun him up yet, but that's, uh, he's coming together. Starting to bring these pieces together. My big challenge being in security. Um, don't even need to be in security to figure this out. But my big challenge is take this company Cluley, for example, I don't know anything about them. I don't know anything about the founder. I don't know anything about the team. More importantly, I don't know anything about their security. So do you think I'm going to have it recording all of my screens with all of my microphones and uploading all of that content to their servers so their eyes can parse it? Right. What if I'm talking to somebody over here about some government thing and someone else shares with me? Hey, you know, we just got breached. Don't mention it to anyone. Someone else is telling me about their life, right? Um, I had a friend die recently. What if I'm talking to my friend about, you know, my other friend who's died? All of this is being uploaded constantly to a third party. Not to Google, not to Apple. Okay. Because. Because I could see doing that with Google or Apple once they, you know, once they sort it out, once Apple sorts out, you know, the successor to Siri or whatever. Google they have rock solid security. I'm not worried about them getting breached or them losing the data or whatever. Other than small issues. But ultimately those two, maybe even Amazon, maybe even Amazon. They also have rock solid security in terms of like how many billions of dollars they spend on it. The quality of their team, the size of their team, etc.. Right. But a third party, a new startup called Cluley and it's all this AI hype is around it. How big is their security team? How much are they thinking about this? Not only that, but how good are they going to be at defending against a halfway decent attacker, let alone a state actor who wants to come and steal the entire thing and see all this context that all these people are spraying up into the cloud. Right? I'm very forward leaning for tech and for AI, but do not go in and record every single thing, every single text, every single screenshot. This thing is screenshot Screenshotting your your displays all the time, all the texture on there. You see on there it could see the app that it's doing. It could see your slack conversations. It sees all this right. So you should be careful. You should be careful about who you're sending this stuff to. I would say if it's Google or Apple, which is not it's not out yet for them, but with Microsoft. Microsoft is another one. I wouldn't worry too much about sending that stuff to Microsoft. I'm sure there'll be problems, but they are a legit security team because they are a legit company. The startups are what you have to worry about, and I'm not saying there's anything wrong with Cluley security, I'm just saying that they are a startup and they're probably not prioritizing security. Um, even if they know how to write, um, show me that they hire an awesome security team, maybe I'll feel better about it, but there's still a startup. Um, so anyway, all that is, um, bringing us towards where it's going, which is? Your Da will be listening all the time. It will have all of your context. My telos file. That is part of what Chi has, right? Oh, by the way, um, Network Truck just released a video about Telos. Just came out, uh, a couple hours ago. You got to go check that out. But that Telos file that he talks about there, that I have, that he has a whole bunch of my friends have. That telos file is like my soul. This this thing is like it's traumas. It's history. It's my journal. It's like my goals, my aspirations, like my problems, my challenges, all this different stuff. Right? All this is available to Chi. Chi knows this. Chi also knows that I like, you know, prawn cocktail. It also knows that I don't sit anywhere where there's like smokers nearby. Right. So it knows all these different things, knows preferences for food, you know, so it can help me, uh, optimize my surroundings. It could order for me. It could do all these things. How is it doing that? By manipulating the APIs that it has access to for the world around us. Right. If I go into a restaurant and I can control the TV, well, I'm not going to control the TV. I'm not going to ask them to change the channel. No, Chi is going to change the channel to table tennis. If I'm not overruled by other people trying to have it change to something else, right? Why? Because this is all available on media. This is part of the API. This is part of the demon for that restaurant. It's just an API key can submit to it on my behalf. And this will be the case for temperature, for sounds, for what's playing on the video, for having doors open for you, you know, for having materials ready for you and summaries created. Most importantly, Chi will be reading a billion different stories and articles and research papers and reports and raw data sets constantly. While I'm in the middle of this sentence and while I'm sleeping, and while I'm in the middle of another project and doing some security work or whatever I'm doing, it's processing all of that and getting it ready for me. The other thing it's going to do is this dynamic content generation, which is a blog post I put out who knows when. Sometime last year Chi will be making me videos. Okay, Chi will deepfake Chuck, who just made that full size video, which is awesome, which you want to go watch the the actual full thing, but Chi will make me a 32nd version of that if I need it. And maybe he just cuts up the video from from Chuck, but maybe Kai makes me a whole new one with Chuck's face or Richard Feynman's face on it, and he can make a video for a blog post which didn't actually have a video. Or he could turn a video into a blog post because maybe somebody doesn't like to watch videos. Or maybe it's audio only because I'm out on a run and I don't want to look at anything and I can't look at anything. Bottom line is, Kai, your da becomes the medium to you. Um, and this this is kind of a big thing. I don't know why this is, uh, becoming a whole episode about this or whatever, but SEO is dead because websites are dead. Advertising is screwed because advertising no longer is going to be landing on the human. The advertising is going to get blocked by the Da. Largely, the Da is going to go and collect things. It's going to be the inbox and the, uh, the inbound filter to every principles life, to every user's life. The Da will then determine what gets through right? And so now the whole attack point for influence against a person becomes their digital assistant. If you want to sell to them, you sell to their digital assistant. If you want to sell to them, you get into the models so that when the digital assistant goes and does its research for what the best you know, pillowcase cover is, it's going to find your product, right. So this is already starting. People are already trying to poison models or influence models, however you want to put it. Um, because they understand that marketing and influence and uh, website stats and all of that, all of that is shifting to go towards the Da. Um, all this content. Nobody's reading this stuff, right? There's too many YouTube videos. There's too many blogs. Like, we already figured that out a long time ago. There's a limit to this, but your Da can go read all that stuff. It can go process all that stuff. And it could use other services which are companies themselves, that are consuming all of that content and producing rankings and directories, which kind of like an old like Yahoo directory or whatever, but constantly updated. And your Da will go review that and be like, oh, turns out that's the best tactical flashlight for right now. Or, you know, that's the best, um, bed cover or the best eight sleep competitor or whatever it is, right? And so what ends up happening is we just we in our Das, our Da is going to be our most important relationship, um, for a great many people, right? Because it could be our coach. It could be our friend. It'll be our therapist. And this is something we have to watch out for because we want our best friend, our best human friends to be our best friends. Right? We want our our husband and our wife and our kids to be our best friends. And we want that to be the center. We want humanity to be the center, right? So we really have to watch this, this gulf that we're about to go into this chasm or whatever, where it's like this thing being offered. This Da is so powerful and it's so awesome, and mine's going to be so awesome very soon. I'm going to have mine kick in long before most people do, uh, because it's already starting. But everyone's going to have this soon. And what we have to guard against is like cutting out actual humans in exchange for this thing that is more responsive. It's nicer to us or whatever. And like, it's going to have advantages over humans. That's fine. But we've got to maintain this is my admonition to everyone. We've got to maintain the mentality of all of this is to enhance humanity. Okay. The goal of AI, um, relationships, the goal of AI therapy, the goal of AI anything should be to take the things that you love about human to human interaction and human to human value exchange and make them better. Find ways to enhance them, find ways to do them more often. Find them ways to do to do them in a more enriched and deeper way. Right. Augmentation versus replacement. Augmentation versus replacement. This is the trick that we cannot ignore. And another way that I really like maximize this is, um, all this infrastructure that I'm building, which, by the way, the name for mine is called Pi Pi personal AI infrastructure. My pi that I'm building, my personal AI infrastructure stresses when a piece of content comes through that I should watch in a slow version. That's analog, that's old school, that's human. Sit down with a moleskine and a space pen and take notes. Slow it down or pause it. Take the notes and think. Maybe call your friend. Maybe have an actual, you know, voice conversation with a friend about this cool idea. I do this all the time with Joseph and with Jason and be like, hey, have you seen this thing? Hey, let's talk about it, blah blah blah. And you know, you don't want to be like, oh, well, that's not good. You know, the the pen is not as good as the typing. You know, the Moleskine is not as good as, you know, putting it in my telo's file. No, this is not a competition. The telos file should not be replacing things. It should be augmenting. Right. It's just a journal. It's just capturing your life. Right. You could do that on a moschino. You could do that in a telos file with vim and markdown. Right. We've got to stay focused on what actually matters and what we value and what we're trying to enhance versus what is shiny and cool and awesome. Right? I'm a security guy. I love security, I love hacking, I love hacking tools. I love all those things, I love I, I love AI agents, I love automation, I love all that stuff. Those are tools. Those are secondary. They serve the purpose of human right. So so I think this is absolutely critical to to just keep in mind, for yourself. But also when you're hearing me talk about how cool these things are, I just realize that they are secondary. And now I have to find my place. Because what, uh, what a divergence there. Um, let's see here. All right. Meta's AI can produce nearly half of Harry Potter one, so we could actually give the actual quotes. I'm curious how this is possible because it's not a database. Um, I don't understand how it's producing exact quotes if it's not a database. Um, yeah, I guess I'll have to figure that out. It's kind of interesting. Um, and I'm not sure I could do that deterministically either. I think a lot of that might be just collapsing on the most likely because it's seen so many samples. Um, but I don't think it could do it, like over and over and over, because again, it's not an actual lookup. It is a generation, which is why you can have, uh, hallucinations and, uh, non-deterministic results technology, Jassy told employees that Amazon will need fewer people doing current jobs and more doing other types of jobs as AI gets rolled out across the company. We got to listen to these folks when they tell us. For one, I'm telling you that this is happening, that they're replacing jobs like whatever that matters. I'm just a guy on the internet talking, but all these CEOs, lots of them are basically saying, yeah, yeah, we're getting rid of jobs. We're getting we're changing the entire company. We're thinking AI first. And a lot of people are like, no, no, they're not going to replace jobs. I can't replace jobs and I won't replace jobs. It's a big, uh, sham or whatever. And I'm like, well, the people who actually control the jobs are telling us this, and you got to watch out because some people are also selling an AI product. So it's kind of like marketing, like I felt that way with Salesforce. They're like, oh, we sell a virtual, you know, AI employee. Oh, and also we're getting rid of a bunch of jobs because we could do it internally. Well, that sounds like marketing. But when Amazon's actual CEO tells you this when? When the head of AI companies like the head of anthropic. I love that guy. Like he's way different than Sam. And Sam's awesome too. But I love how human focused that the head of anthropic is. I love how he's like, people get ready, this is going to take jobs like it's going to suck, but it's going to potentially be awesome. But get ready. Why would you not listen to that? What do you think he has access to? What do you think he's looking at to make him think that? Right? This is not a stupid person, right? So just when you when you see these CEOs that have no reason to lie to you. And I hate to use this analogy, but when a terrorist tells you I hate everything about you and your culture and I want to destroy you, and if I were to blow myself up and you were to die as a result, I would go to heaven. The one thing you shouldn't do is be like, yeah, I'm just not really sure he believes it. No, he believes it. He believes it. And you should be careful at the park as a result. Okay. And this is, like, the worst analogy ever. It's like the worst analogy ever. because I'm not saying right. The heads of these companies or the people making these models are terrorists. I'm just saying that when somebody is crazy about a belief and they are like telling you from the depths of their soul that they believe it and it is true and it is going to happen, it doesn't mean it's going to happen, but it does mean that you should believe them, right? You should believe them that they believe it. Again, with the caveat of of marketing. Meta's expanding beyond Ray-Ban with Oakley glasses aimed at athletes. I'm getting a pair of these. My first pair was fantastic. I can't wait for the battery to get much better. Resolution is improving. Apple, where are you? We're looking for you. Gartner says I will handle half of all business decisions by 2027. Um, did I miss something that is way too fast? Way too fast. It doesn't matter if we get asi si, not agi if we get ASI next month. Okay, so it's a 400 IQ or a 4000 IQ. It's smarter than Albert Einstein and it runs on your phone for $0.03. It doesn't matter if everyone had access to that, it still wouldn't have half of all business decisions in 2027. It doesn't matter how smart the AI is. Companies move slow for a reason. There is friction built into the system. You can't reinvent yourself. You can't reinvent the leadership team. You got regulations to go by. You have employees like there's way more friction in these companies. And and I'm saying this with the experience of like working in like fortune ten companies for years at a time. You know, being at Apple, being at, you know, anyway, I don't need to describe all that. Uh, trust me, companies move slow. Uh, there's nothing that's going to make something handle half of all business decisions by 2027. And maybe that was just a bad, uh, take on what was actually said, but I don't think so. iOS 26 opens up AirDrop and Airplay tech. Two third party apps this is all part of. If you haven't been paying attention to why Apple is opening everything up. It's because of antitrust. They are very afraid of getting crushed by antitrust, so they are opening up the entire ecosystem. That's why you're seeing this. And iOS 26 finally lets you set custom ringtones the easy way. So that's cool. Humans. It looks like we found all the missing matter in the universe. Did you know that we were missing, I think 80%. Was it 80%? I can't remember the percentage. We were missing most of the energy in the universe. Like the the calculations, the equations, they do not add up because basically, I think this is correct. Um, if we only had the amount of matter that we have seen, then the universe would be expanding a lot faster. It's not expanding faster, which means there's a lot more gravity holding it in, which means more matter holding it in. And we have been unable to find that matter, but we just did. So they use these pulses to find and essentially it is all in this fog, this very light, very thin matter, fog between galaxies. And that fog altogether makes up the other large percentage that they were looking for. And this is almost guaranteed to be a Nobel Prize in physics. This is like one of the biggest findings ever in all of astronomy. And huge, huge news. Also, same week Rubin Observatory takes its first images. Absolutely stunning. I'm getting a print of this thing. It's like the Hubble Deep field image, if you remember that. Well, that just times a million. This thing is insane. This Rubin telescope is different and better than anything else that's ever been put out there. I've got like, 20 links to this thing and like all the images and the story of it. YouTube video. You got to go check this out. Now, they seem to have been brought online to investigate dark matter. I wonder if they were trying to figure out the the, uh, the secret of dark matter and actually find it. It would really suck if that were the case, because they are announcing their existence the same exact week that the whole mystery and puzzle was solved, so. Holy crap. If I'm right about that, the timing has to be like a little bit of a letdown. But still, the images are just insane. And they're doing it with like, multiple mirrors and a super, super sensitive sensor. MIT researchers using EEG monitoring to track brain activity in essay writers and found ChatGPT users show the lowest brain engagement. This is basically saying AI makes you way dumber. Um, I think we talked about this last week. I don't think we need to go into it too much more, but sample size was fairly small, but it absolutely showed that if you rely on AI too much, it will make you way worse. The way the way I think about this, the way I explain this is, is, I think, fairly elegant. If you were the type of person like me who is going to use Chi Chi's job, my DA's job is going to be to constantly question me, to constantly harass me, to constantly debate me, pull out my bias, challenge me, force me to like, make a statement speech. You know, um, like I'm on an NPR debate or IAC squared debate. I've got to, like, perfectly defend my position. Okay, now switch to the other side and make and make their choice or make make their argument better than they made it. Right. I'm teaching Chi to be an augmentation system, a training system to make me better as a human, to make me a better first principles thinker, and to make me more independent of a thinker. Right. I is going to enable that 100%. It's going to enable that for millions upon millions of people, hopefully hundreds of millions of people, hopefully billions of people. That's the good scenario. The bad scenario is you just use it like a crutch. And that's what I believe we're seeing with this study where people got way stupider. Right? Oh, the most interesting part about this. They actually saw brain activity decrease. Okay. So they have physical evidence of, I would say, just people just being dumber, like their brains are not as active because they are offloading that stuff to an AI. And that is super frightening to me. That is like really, really a problem. Just keep in mind it doesn't mean it's it's necessary. It doesn't mean it's guaranteed. Right? You can actively defend against this and have it actually be a net positive. It's a choice. Massive study analyzing 1.2 million dating app ratings found men rate women as attractive 13 times more than women rate men. Calling it the attractiveness gap. Women rate only 4.5% of men as attractive. That's quite sad. What gets measured. I will automate. Uh, you could check that one out. Microsoft data shows the infinite workday has become the norm. Analyze trillions of data points and found that work now stretches endlessly, with employees getting interrupted every two minutes during business hours. Yeah. Yep. That's why I got out of the game. 2024 Baby name details. Extreme spelling is exploding. Ten of the top 20 Rising Boys names have X's or Z's in them. Finland is firing up the world's largest sand battery. Evidently store a lot of heat and sand. This is like tripping me out. Private equity has taken over and destroyed much of America. Megan Greenwald's new book argues that private equity firms have systematically bought up the entire industries and optimized them into bankruptcy. New Covid variant, called NB 1.8.1, jumped from 15% to 37% of US cases in just two weeks. and people are worried about a potential summer surge. What was interesting here is there's actually CDC data on this, which I'm surprised and happy that there's still data coming out of the CDC discovery. Again, I'm not going to read all these. They can get quite long. And there is the newsletter for you to go check out. Heartwarming. I, I am going to read this one. So a teacher shows her students eye images. This thing is insane. You got to go watch this video. It shows them. So somebody's like, I want to be a football star. I want to be, you know, a marine biologist. I want to be an astronaut. It shows them age accelerated versions of them, like in an astronaut suit or whatever. And just like the looks on these kids faces, it is so, like, so wholesome, right? Just imagine seeing that it is possible, right? Do you believe your parents when they say it? Okay, a little bit. Do you believe your teacher when they say it's possible? A little bit, but how does that compare to actually seeing your face grown up in an actual astronaut suit? It is extraordinary. You got to check it out. Good workers are often bad at interviews. Hormozi is highest ROI spins. There are tons of great discovery links. Again, I don't want to just go through and read the title. Go check it out. Okay, this is the end of the standard edition of the podcast, which includes just the news items for the week to get the rest of the episode, which includes much more of my analysis, the ideas section, and the weekly member essay. Please consider becoming a member. As a member, you get access to all sorts of stuff, most importantly, access to our extraordinary community of over a thousand brilliant and kind people in industries like cybersecurity, AI, and the humanities. You also get access to the UL Book Club, dedicated member content and events, and lots more. Plus, you'll get a dedicated podcast feed that you can put into your client that gets you the full member edition of the podcast that basically doesn't have this in it, and just goes all the way through with all the different sections. So to become a member and get all that, just head over to Daniel Store.com. That's Daniel Miessler, and we'll see you next time.