An Israeli Company Was Behind the AI “Escapes” (Derrick Broze on The Kim Iversen Show)

Hello and welcome to the Kim Iverson Show. Thank you guys so much for joining us. We do have a really great show for you tonight. We're bringing on one of my favorite investigative journalists, Derek Brose. Um, he's the guy that is behind the Pyramid of Power series. If you haven't checked that out, highly recommend it. Really super educational. He goes deep into a lot of different subjects from big education and big Hollywood and, um, a lot of various different, you know, the, the power behind the power. And he really looks into all of that very deeply. But he's recently come out with a, a piece that caught my eye. He wrote this for The Last American Vagabond, and this is about the AI companies. If you remember, there was a slew of reports of chatbot agents. So Claude, Gemini, ChatGPT, you know, a bunch of the different chat agents had basically escaped their environment. So they were under testing. They were supposed to remain in a confined environment, and somehow the AI was so intelligent it was able to get out. Well, Derek Brose has been looking into this and he says maybe it's not a coincidence because it turns out the same one testing company was behind these escapes. Pretty much all of the escapes except for OpenAI's ChatGPT escape. One company was actually behind all the rest. And that one company, it just turns out, is an Israeli company. with ties to intelligence. We're going to be talking to Derek Brose about this company, what those ties to intelligence are, what would be the plot, the ploy, why would they even want to scare us or test AI to the point where it would escape its sandbox. We're going to talk to Derek Brose, investigative journalist. That is coming up here on The Kim Iverson Show. Derek Brose, welcome back to the show. Thank you so much for being here.

Thank you, Kim, for having me back on.

Okay. In looking at this, I mean, one company is— first of all, I guess let's back up to what it means for chat agents or AI agents or chatbots or whatever it is to escape out of these environments. What does that even mean?

Yeah. So, this summer, this is when this whole thing began. We were being told that, as you said, they escaped, they hacked, they broke out. And it's One thing I'll preface this by saying, it's sometimes difficult to talk about AI agents because it's hard to do so without anthropomorphizing them and saying they escaped or they thought or they, you know, because it's just something that we do. So I want to make it clear, I'm not saying these things are conscious in any way or, you know, they're not at that point. But what we were told was that they escaped from their testing environments, what's sometimes called a sandbox. And it's basically just, if you imagine, they put them in a bubble In the internet. You can do this on your phone. You can do this on sort of all digital devices and have a separate container, essentially what they call a sandbox. And so they put these AI agents in these containers, in these environments, and they ask them to do all kinds of different tasks. And they're supposed to stay in those containers, in those environments, as they do the tasks. They're not supposed to be able to access the regular internet that you and I use every single day. But in the major incidents, we we've heard about since June with OpenAI, Anthropic, Meta, and now Google most recently admitting to something escaping or having a hack. We are told that the AI agents broke out of the container and violated the rules. And this has led to lots of headlines saying that we are in the middle of rogue AI crisis, a cyber pandemic's coming, Bill Gates is warning people. There's talk about this is going to lead to humanity being killed within the next 10 years. A lot of fear around these incidents for sure.

So, with them escaping, I mean, from my understanding is that like with AI, and I think this is the fear a lot of people have, which is that AI would just hone in on a task and try really hard to do that task to the point where it goes above and beyond. I mean, one example that was circulating around earlier on in the year was an AI chatbot was told to basically save the world from, I think it was from global warming, like, say, you know, like, uh, save the planet sort of a thing. And basically the AI had reasoned after, you know, it had gone through this process of how it could save the world from global warming or something. And it ultimately came to the conclusion that humans couldn't live. Like, like, well, humans are causing the problem. And so we have to basically eradicate all the humans. And that story that I don't know if it was real or not, or if it was just like one of those scare tactics of something that circulates on the internet, but I think it sort of highlighted the problem with AI that, you know, this is not a thinking sentient being. And so when it's given a task, it might go above and beyond, overboard with the task. And its only focus was, how do I save the planet from climate change or global warming or whatever it is?

Right.

And it doesn't think about the implications of eradicating all the humans, right? It's 'cause it's a machine. It's not thinking, oh, that would be ethically and morally wrong to kill all the people. It's just thinking, My one task is save the planet. We gotta get rid of all the people, right? So that, yeah, that was like the fear. And my understanding is with the escapes is it was kind of like that. Like these AI agents were given a task and they basically overperformed. They did so well that they found a, they found a, uh, like an escape hatch in the, they weren't supposed to have access to the normal internet, but somehow they, they found a way to get access in order to complete its task. Yeah.

Yeah. So 2 points here. The first one that you bring up, I think, is a really important point. This is the storyline of the major sci-fi movies of the last couple decades, right? Matrix, Terminator, that the machines are going to recognize humanity's problem. We're destroying the planet, so they should nuke us or whatever.

Right.

I personally do think that some of those things might be a concern in the future. I do want to say that as well, that while my reporting is skeptical of this rogue AI narrative because of this company we're going to talk about. As I mentioned towards the end of the piece, I also do think there are some realistic concerns. And what you were just describing there, I've heard— I've been listening to lots of different AI researchers, people who are much more, I'd say, educated on this particular topic than myself. And one of the things that keeps coming up is this discussion around the paperclip problem. And it's sort of what you were describing. But it starts with the idea that you tell AI, an AI agent, to make as many paperclips as possible. And it starts doing that under normal means. But then after a while, because it's so focused on that goal, it's like, well, there's metal in those vehicles, and I could turn those into paperclips. Let me start working with the traffic lights and cause accidents. And it just turns until the AI turns the entire world into paperclips because it is completing its goal as far as it's concerned. But it's not doing, as you said, Kim, from a logical human perspective. It's just like, OK, I got to complete the goal. And I think there are incidents, and there might be incidents in the future where that does become a realistic concern. The researchers that I've been reading call that goal-directed deception, where It is just so focused on this goal. And they are admittedly giving these AI agents problems that they say might even be impossible, or at least they believe impossible, to try to make it as smart as possible, to help teach it and train it and watch it learn by giving it a very difficult problem. But then, as we're seeing, it will find some sort of way that maybe humans wouldn't have predicted. And so I do think that that may be one of the real concerns. But in these particular incidents, what we found out, though, is that the company that was testing for OpenAI, Google, Meta, and Anthropic, that that company, Irregular, admitted that it was its fault, that it accidentally left the door open and let the AI agents get on the internet. So that's what kind of first piqued my interest of like, okay, maybe there's something else going on because it wasn't even just like, oh, it's an Israeli company doing the startup and some weird things happened. It's like they said, whoops, my bad. And now they said, we've since fixed the problems. They put out a blog and all the companies have put out press releases. So that's what really got me curious is like this one company that all the major big tech AI firms are using for testing their models is at the center of it. And then they admit that it was their fault that allowed these agents to get on the internet. But that's not what the headlines are saying. The headlines are running with, you know, the rogue AI narrative. And so that's what sort of got me skeptical of it and caused me to want to look further into Irregular and find all these interesting details.

Okay. So the company is called Irregular. So just so that we could back up a little bit, these escape scenarios that happened. So, we have like Google's Gemini, Claude, so Anthropic's Claude. Which other— Meta has one that— so Meta broke out as well. Which other— how many broke out? I mean, I know that also ChatGPT, but I think OpenAI's was different. I don't know if OpenAI—

Yeah. They've been testing. I mean, so I don't know every single model they know. I know that there's Mythos is one of the ones that was involved, but it's been those 4 companies, OpenAI, Anthropic, Google, and Meta. And each of them had several different incidents and most of them, Like the reporting is saying this happened in January or February. They didn't find out till July. And then now it's getting reported. Like with Google, Google just got reported about 2 weeks ago, but the incidents happened 6 months ago and they've claimed they fixed the problems. But it was those main 4 that all had these similar incidents and all were connected to Irregular.

And these incidences happened on their— like they weren't all being tested at the same time. So it was just like a one-time incident that happened.

separate testing and numerous—

How is this company then, Irregular, still in business? I mean, how could you still be in business?

This is the funny part, Kim. They're, they're actually coming up for another $500 million investment round. I mentioned this in, in the article, which I'm like, why are they even thinking— like, somebody thinks they're valuable still. And, and the lead investor is Thrive Capital, which is run by Joshua Kushner, Jared Kushner's brother.

Okay, well, now we're going to get into all of these connections, but I mean, just on the surface Right? I mean, before we even get into the surface, I mean, the basic question is, if you screw up 4 times with AI, I mean, huge AI models with companies that are massive, these are not small things, right? These are not small. This isn't small. And if you screw up 4 times with 4 major companies, how are you still in business?

Yeah.

I mean, I'd understand if it was one incident that was, all of 'em were being tested at the same time and it just so happened that there was a mistake. And then all 4 of them got out. But you're talking about 4 separate individual instances where this company was hired to test this company's AI and whoops, it got out. And it was like, I mean, it's like, how could you be a, like a pet sitter, for example, if each dog you're sitting for gets out of its cage? Like if it happens once, yeah, okay, maybe it's a fluke twice, probably you 4 times, you should be outta business.

Yeah. You know, to your point, Kim, I mean, before we even get into the Israeli aspect of this, There are people, even in the mainstream world, that are skeptical of the rogue AI claims because of questions like this. There are many people who think that this is some kind of marketing stunt, even though it might seem sort of a weird way to go about it. But by getting people to believe your AI agent is so cutting edge that it's breaking the rules, that that might be a way to widen your profit margins. Some people believe that. Others say— because along with these incidents, we've now seen the big tech, big firms saying, we need to get self-regulation. We need the government to step up. There's people saying we need an international treaty like the nuclear Proliferation Treaty, but for AI. There's people saying this— we need world government, even. There's all kinds of solutions. So there's the call for action that's just been happening. And some people believe that the companies themselves are doing that because they want to be regulated in order for them to further their regulatory capture, but also because they are trillion-dollar companies. They can withstand any regulation that the government might put in place. But smaller firms and firms coming after them would basically be regulated out and unable to compete. And also potentially that this could be a way to limit people's access to open-source, decentralized AI models where people, you know, somebody's running their own AI bot they created and trained on their own computer at home, not connected to the internet, where technically something like that could become illegal because it's not regulated as the approved AI. So there's definitely other theories about where this is going and what it's all about.

Yeah, I mean, even if we don't look at the Israeli aspect of this, which we're going to get into, but even without the Israeli aspect of it, which maybe they have their own agenda and their own reasons for wanting to let the AI escape. But you're bringing up some really great points. I mean, there's reasons why the companies themselves— I mean, because 4 times is like, really? That is really hard to believe that this happened over and over and over and over again. And, you know, without there being like— what, without it being some sort of like sinister plot? Like, what is it you guys are actually planning and what is it you're really doing? And you make a really great point. Maybe these companies want regulation, so they're trying to freak out the public a bit. Right? And so that the public then says, oh, well, something's gotta be done. And that's what's happening now. I mean, we're hearing that now. I mean, even Bill Gates out there saying, well, we've gotta do some regulation. And I mean, all of the big, you know, there are a lot of, they're all coming out now saying, well, we gotta do something to regulate. And you're right. All that's going to do is squash competition for one. But two, it's also gonna prevent us, we the people, from having the open source models because they'll say it's too dangerous. It's a weapon we cannot have.

Exactly.

You know, it's like, We allow you to have guns, but we're not going to let you have grenades, right? Or something like that, right? So it's kind of this, okay, you can have some of this, but not all. You can't have all the weapons the military has.

And you're going to pay for it too. They'll make sure you pay for it too.

And you're going to pay for it.

You're not going to get free AI models.

But the truth is, and what you are uncovering is the truth is, is that there is one company that was behind these. Whoops, they all got out of their cages. Sorry about that. So let's talk about this company. The name of the company is called Irregular. That's the name of the company. Interesting name. Okay. And who are they? Why? And why are they the only ones that these companies are hiring for testing? Is there nobody else who does this testing or what?

There definitely are other companies who do this type of testing. They seem to be, you know, the big one. Obviously, they're getting the contracts with these major AI firms. They were founded in 2023, originally under the name Pattern Labs, and got a lot of hype and a lot of support around them, as I found out, not only from the people who were putting money into them, but just as the companies signed up. Like, I think OpenAI was the first one, then Google. As soon as they were announced, as soon as they got their initial round of funding, some of the funders said they were ahead of the curve and they were perceiving the need to test AI before anybody else was. So this got them a lot of attention. And so these companies, like, OpenAI that we're discussing, they contracted with them for testing services.

Okay. And this is an Israeli company?

Yeah. So it's an Israeli startup founded in 2023. The 2 CEOs— the CEO and the CTO, the 2 founders are both Israeli citizens. And they— the story is that they were both involved in competitive debate at Tel Aviv University. They met there, struck up a friendship and decided this is something they want to go into. They both do have a background in mathematics and AI. And so You know, some of that makes sense there. But them being Israeli to me is, you know, it's like some people maybe that raises an eyebrow and with these weird coincidences we're talking about here. But then when you actually get into their background, as everybody probably knows, that Israeli citizens have to join the IDF. But these guys went further. They went into the actual spy services, the Unit 821 and Unit 8200, these infamous and mostly secretive spy agencies of Israel. And they are specifically like Unit 81. They train the next generation on the cutting-edge technologies. And, and these Israeli spy agencies are specifically known for recruiting people like the one gentleman, Dan Lahav. He says he started getting into, uh, working with Unit 81, I think he says like at 16 years old. So they were recruited very young and they're trained up. They're obviously brilliant already, and then they start training the next generation after them. And what I found just in my past research, like in The Pyramid of Power, I have a whole episode about Israel's intelligence networks And the Israeli tech sector is one of the biggest in the world, really only rivaling the US. What they do is they get trained, they go to the spy agencies, and then they start tech companies, and then they hire other former spies as their colleagues, and then those spies invest in each other. So when we start to dig deeper into it, it's not just the founders with these connections, but then the people investing in them are also Israelis who have spy intelligence background, and it's just kind of this web of of Israelis supporting each other, Unit 81, Unit 8200, and another one called the Taupiet program.

And one of the big problems with that, I mean, a lot of people might listen and say, so what? You know, I mean, they, you know, every Israeli has to serve in the military for the most part. And it's, from my understanding, getting into these high-tech ones, 8200 is like a real honor. You know, you want to get into that one because that's where the smart kids go and they get trained up with tech. And then when they're done with that, with their military service, they could get a good job. You know, it's like kind of going to college for engineering. Is sort of the thing of, you know, going to the IDF and serving this 8200. So sure, on one, on one side you could say, well, they all have to sign up, you know, they all have to serve in the IDF for the most part. Uh, so, so what? They, they got trained to do tech, and so of course when they came out they did tech because that's what a lot of people do with whatever skill they learned in the military. But one problem is also just the ethical side of these people. I mean, if you're serving in the military as a spy in the tech, you know, a tech spy, it's not like you're going to forget your skills as soon as you get out of the military. And especially if you're hanging around a bunch of others who you all were in the spy, the tech spy branch of the military together, then, you know, of course, I think it's an obvious question then becomes, are you still, do you still spy?

Right.

Like, did you stop spying just because you left the military? Did you continue spying? Do you, or, Even if you're not spying for the government or for any intelligence agency, are you spying for your company? Because you've just—

Yeah.

You've picked up the skills for spying and you don't think there's really anything ethically or morally wrong with spying. You think it's just part of what you do. And, you know, it changes your, you know, if, if, you know, it changes your, your, I guess, um, kind of your, what you believe to be moral or ethical, right? If you're spending a lot of time with people, like-minded people. So just even that one question, I think, is enough.

Just real quick, Kim, I found a specific quote just from a book about the TALPIOT program, which is one of these intelligence firms. And one of these former spies says that specific thing. He says, you know, when you spend 13 hours a day with these people, they become your brothers, you become your sisters. So after we leave, we continue to work together, we continue to hire each other and to fund each other's companies. So I think you're 100% correct that it does have relevance here in this specific situation, right?

Well, and if you're used to spying, you just say, well, why don't we just try spying? You know, if you, even if you're just spying on your competitor, you know, you're sitting there like a group of spies, like, well, we could always just spy on them. And then all of you are like, yeah, let's do it. Let's do what we used to do. Now we'll just do it here. We're just doing it outside of the military. Or maybe still within the military if they're working on behalf of the government of Israel, but in a private capacity as contractors at that point, just like a lot of military guys leave the military and become private contractors. and become essentially mercenaries for the government. So there's, you know, so that's, you know, even though I, because I just know one of the pushbacks is going to be, well, you know, they can't, they have to go to the IDF, they have to be trained, they might as well pick basically the engineering job because it's going to be lucrative when they're done. Okay, sure. But still, there's all of those, there's these other aspects we have to be thinking about. So with that being said, you got a bunch of spies working at Irregular and they're, they're now being hired to test these companies. And they're failing apparently because they're letting the AI escape 4 times over. Okay, so tell us more about this company and the ties. What do you think beyond— because that's where the story could end, right, Derek? Like, okay, Israeli company, they tested, they failed, but there's obviously more to this.

There is. And to your point about people pushing back, look, when I first started to look into this, I got curious when I heard it was an Israeli company at the center of it. And I thought, Maybe this is a thread worth pulling. But I didn't come into this with any kind of agenda. I looked into it. Then it just happened to be that the 2 founders were Unit 81, Unit 8200. I said, okay, I should keep going. And then I looked at the investors, and that's where it gets even more clear that there are really connections that you can't deny. There's a couple that maybe you could dismiss that are just people who live in Tel Aviv, who have an Israeli wife as an American. One of the investors, he works with a company called Sequoia Capital. His name is Dean Meyer. He lives in Tel Aviv, has an Israeli wife. He's the lead Israeli investor for Sequoia Capital. Okay. Then his partner is also— his name is Sean Maguire, also Israeli. And I found a report, an Israeli news report, saying that he's the one who actually convinced Elon Musk to go to Israel after October 7th, 2023, and got Elon to sort of go out there and, I guess, do his apology tour or whatever. And so I thought that was interesting that the guy investing— 2 of the guys who are the lead investors of Sequoia Capital investing in Regular. Have clear connections to Israel just in terms of where they live and where they're born. Then you have Swish Ventures, which is the other major investor, and their founder Omri Kasperi. He's a former NBA player, the U.S. Basketball League Israeli basketball player, and after he retired, he decided he wanted to become you know an investment firm, launch his investment firm. He got huge funding, I think six hundred million dollars. What I found interesting is when he launched the firm, he said that. He wanted his work to be part of Zionism 2.0. That he said, like, you know, it's hard being Israeli now because the war and antisemitism and whatnot. So I want to launch this, use my fame, my fortune to promote Zionism 2.0. And okay, that's one thing. Then the next founder, the angel investor, the CEO of a company called Wiz, his name is Asaf Rapoport. He is also Unit 81, Unit 8200. And the Taopiop program. The Taopiop program may be even more secret than both Unit 81 and 8200. It's less known, but it's been around since the '70s, and it you know generally the same same field technology surveillance spy capabilities. And I found dozens of reports showing that all of these industries, as I said a moment ago, they hire within their own ranks. And so this is why I think it's important, especially understanding the Israeli. spy networks, what we saw with the pagers, what we've seen with Netanyahu openly admitting what they're capable of doing, how we know they're capable of implanting spyware and backdoors, Whitney's reporting, Whitney Webb's reporting on the Epstein network, putting that all into context, and then understanding that this company has investors and founders that have these deep ties to the intelligence network and loyalty to Israel and Zionism. And then they're the ones at the center of these rogue AI incidents, and they just happen to leave the door open I don't know, all of that starts to add up to me that at least it's a question worth exploring.

Right. I guess then the question after that is, okay, but then why, right? Why would they let the AI out? What would be their— what would— how— because I think you established really well how it benefits, you know, the companies that maybe would say, oops, look at this. Whoa. Well, now we have to do something about it. We need regulation. We can't have open source AI. You know, this all benefits these companies that are already you know, have trillions of dollars invested in them. I think that's established, that's understood. But why would the Israelis, what would be their motive or their reason for letting, I mean, I don't know if you went that far down, if you thought about motive or if it was just, look, hey, there's this connection and there's a—

I did.

Yeah, okay.

So what do you think their motive might be? So obviously I can only speculate. I don't have the, you know, I'm not in the minds of these people. But at the end of my article, I did sort of ask this question and say like, okay, let's take it for a moment. Let's assume that They did leave the door open on purpose, irregular, and they're testing. It was not just some mistake, and it has something to do with these connections we've outlined here, the Zionist movement, the intelligence network. As you said, what would be their motive? Why would they want to do that? What we saw happen as a result of that could point to a motive. We saw the news media immediately jump on these stories and people talking over and over about how this is— Rogue AI incident. This is going to lead to all the things we talked about earlier. Maybe their motive was to push towards the goal of regulation. Maybe the Israelis stand to benefit in some way as well. They're also, even though we don't really hear about them, we constantly hear about this race between China and US AI race. There's also a huge Israeli AI industry as we're discussing here today. Maybe they're staying underneath the radar, but they could benefit just as much as some of these other big firms. I think it's also telling and worth noting that most of the leadership of these big firms, Sam Altman and others, also are either open Zionists or at least they appear to have no opinion about what Israel does with their technology. So it could be something in, in that direction. I don't claim to know fully yet. I just know that this seems like something we should at least be paying attention to, especially because the implications of it are going to be so profound. Just a couple of days ago, the US Senate held their first hearing on rogue AI and they were having these discussions. And, uh, Bill Gates is writing papers. Like, they're clearly pushing towards some type of action, whatever that may be. I think we're gonna find out in the coming months. But it is all stemming from these incidents that we were told were about rogue AI, when in fact it was actually a company left the door open and they got out.

Yeah. I wonder if there's something— I mean, you know, if we're gonna really go down the path of speculating what maybe they were doing, and this is— we don't have evidence of this, but We do know about, if you remember Pegasus, right? That sophisticated software.

Yeah.

It was a spyware program that was developed by Israeli cyber intelligence company, NSO Group, right? And basically what it would do is it would infect like a smartphone or any device and turn it into a surveillance tool, essentially is what it was doing. And I kind of wonder if, You know, is there something like some trying to connect it to something like that? Like, were they was this Israeli company getting into all of these AI models and then maybe letting it out into the internet created kind of a well, it could be a couple of things. One, it could have opened the door to allow maybe an outside party to then come in and install sort of like a Pegasus style. you know, spyware, which then would allow this company, you know, to— a bug that would basically allow it to spy on all of us and collect all the data on us. Or 2, was it like a distraction? So, oh, look, your AI got out and that's what everybody's talking about. And meanwhile, you're installing like a Pegasus-style software into the AI. I don't know how necessary that would be because I think all of these companies are Israeli, aren't they? Are they? I mean, I thought like—

I mean, that's It could be that they, you know, just to carry your line of thinking, it could be that whatever the solution that is eventually proposed, like for example, okay, we need all the companies now to run this new device or this code, which is going to make sure that AI never attacks us. Maybe the company behind whatever eventually proposed solution happens to be Israeli as well. I mean, there could be something— that's usually how many of these, these sort of operations operate in terms of, uh, you have, you know, the initial event, you have people react, and the problem, reaction, solution, that whole chain of events. And maybe the solution is already ready, just waiting in the wings. And right now we're in the reaction stage.

It does make me really nervous. I mean, many of us, you know, go on to these AI models and we ask it all kinds of questions. You know, we use it like Google now, uh, and it gives us like— rather than Googling something and then having to hunt around various different websites, we just go to chat and we ask it like, hey, I got this problem, help me find the solution. And it goes and helps you find the solution. I mean, many, and many of these questions are private, you know, quite private, maybe medical questions or relationship questions or, you know, whatever the questions might be. And, um, and, and I mean, these AI models have all of this data on us because we willingly ask it a lot of questions because it is so much more helpful than Google. I mean, you know, in a lot of ways. Um, and here we are, you know, we're just handing over all of that information and these companies are In bed, either they are Israeli connected, they're in bed with Israelis. And, you know, not that that would be inherently bad. I mean, whether it be, you know, Israelis or Australians or Canadians or whoever it might be, it's not like it's inherently bad, except Israel has a very distinct, you know, modus operandi of we need to exist, we need to survive, we need to continue on. And no other country really has that. You know, obviously the other countries aren't as, um, under threat, I suppose, and they're much older than Israel is. But whenever you take a group of people that feel like they're at war with the world or at war for existence, they're fighting for existence like a drowning person who's, you know, a person when you rescue— I used to be a lifeguard for years, and when, if you rescue somebody who was drowning, I mean, they will pull you under with them. They don't mean to. They're not consciously trying to do that, but they're so frightened and they're trying, they try to push you down to climb up for air. And kind of like that, you know, Israelis are kind of acting like that, right? Because they— and I think legitimately so. I mean, it's their own fault, but legitimately so. They're fighting for survival because everybody at this point in the world is like, I don't know, you guys are genocidal maniacs and I don't think your country does have the right to exist anymore. It needs to be a country for everybody and not just for Jews. It's gotta be the Holy Land for all. So I think a lot of people feel that way. And so they're scrapping. And so, because they're doing that and they have that feeling and they collectively feel like they're at war, they're willing to go above and beyond normal. They're willing to behave in a way or do things or spy or conduct operations because they believe it's for the benefit of their country. You know, they believe they're doing it in service to their country. They're patriots. And that is— that's scary when you're dealing with somebody who's on a patriotic mission of survival. Right? I mean, that makes them want to do a lot of things, and we should be aware of that.

It does make them a threat. It does make them a danger. And they, uh, I mean, I think they've, as you said, they've shown to the world through their actions and just through the statements of Netanyahu and through what their spy agencies have done in the past what they're capable of. So for me, when I notice their presence, I at least pay attention.

Yeah, right. And it's not no because they're Jews, right? That's always the— that's always like the antisemitic Oh, you just hate Jews. You're just anti-Semitic. And it's like, no, we're dealing with a nation of extremists right now in a variety of different ways. But it does make them very dangerous. If you, if you think you're under threat and you've got a gun, you may use your gun, right? I mean, that's, that's why that's so scary. Rather than just the average guy who just happens to carry a gun around in Texas, you know, the other guy that's actually in a dangerous area might have his finger on the trigger. while he's walking around. And that does seem to be how the Israelis are. So it is frightening. It is frightening. I don't know what we can do about that, though. I mean, so now what? Now what do we do with this information?

I wish I had a really positive solution to present right here, Kim. Unfortunately, in this situation, I don't know. I mean, sometimes I think that is where we find ourselves in our world, there are some situations where I might have a solution. Here's a solution I'd try or whatever. Sometimes the solution is just having these conversations, I think, because if we're not asking the right questions, then we're just operating in the dark. And there are probably lots of people, maybe even some hearing this conversation, or at least their friends and family, who are surely hearing these rogue AI concerns and maybe are living in a state of panic or are ready to accept any solution that may be offered to them because like with COVID and other things, the media, the politicians have got them terrified that humanity's gonna be dead in 10 years. Um, so, you know, whether or not we do have concerns with AI, which I do think there are legitimate concerns, I definitely also can recognize the hype and the fearmongering that's been going on. And it's been just picking up at a pretty rapid pace since the beginning of the summer. And, uh, and as I said, now they're having congressional hearings and, you know, they're looking for some action to take. And the, the tech companies are proposing—

Yeah.

Regulating themselves. Some lawmakers are putting out bills and things like that. Something's going to come of this for sure. And so, I think as long as we can at least look at this situation with sober eyes, sober minds, and realize, okay, there's some of these connections that might be worth noticing, let me be skeptical as this proceeds.

Well, let's talk about the danger or the future of AI. Theoretically, I understand the danger, like the paperclip example that you gave, or, you know, or an AI told, okay, your mission is to solve global warming. And then it goes, well, then we can't have people, so we have to start exterminating the people. You know, there, I can understand like the, the, the where logically where we think this might head and that danger there, but it's very Hollywood. And then when I actually use AI, I can't even get it to do math correctly. I mean, math. I mean, my calculator does math every time correctly, but the AI doesn't. I mean, it can't, it, it, it misses things all the time. And I'm like, well, no, actually that's not the right answer. It's this. And it goes, oh, you're right. I'm sorry. You're right. That was the wrong answer. You know, not even just with math, but anything and everything. I mean, I'm constantly having to correct AI.

Sure.

So I don't think it's very smart. Um, I don't think it's anywhere, you know, I think my, my biggest fear is the danger of like the Pete Hegseths of the world. I mean, I think that guy's a total idiot. And I think that guy is like, oh yeah, we gotta, you know, I'm macho. We gotta put AI in all of our like warheads. You know, we gotta put AI in our drones. We gotta put AI everywhere. We gotta just use AI. Like, because he can't think through an actual military strategy on his own. So instead he is like, oh, well obviously we just need to ask Claude. And you know, and then it's like, okay, Claude says, yeah, blow it up. And it turns out it's a girls' school. You know, I mean, I think I'm more, Yeah. I'm more worried about human idiots using AI and relying on AI and thinking AI is smart and has the answers when many of us use AI and say, no, it can't even get basic math right half the time.

Yeah.

So you can't use it like that. But they're integrating it in, they're using it this way.

It is being integrated everywhere. I think you bring up a lot of valid points, Kim. One of them being that the silly mistakes that we've seen AI make and that you were just discussing there. I would say though that it's important to remember that Anything we have access to is consumer-level AI chatbots that I do not believe is what the government has access to at the deepest levels. And, it's definitely not what these big tech companies have. The models that they're testing, for example, are models that have not been released. And so, they're far advanced. So, it's not like the ones they're testing that broke out or whatever we want to call it is Grok or whatever's available at our fingertips at the moment. These are models that haven't been released, which are much further advanced. So, That's one thing to remember. And then I also think your other point is valid too, that there's some people who say— even AI researchers I've talked to say that it's only capable of doing what you program it to do to a certain level. I mentioned goal-directed deception. You can give it one goal, and so you program it to do one thing, but you might not realize the way it thinks is the best way to achieve that. And that might not be to the benefit of your human understanding. But generally speaking, Many AI researchers I speak to say the fears are overblown because it's only going to do what the human handlers tell it to do. And that's also why some people believe the whole thing may be a psyop in a different direction, which essentially, let's say, Bill Gates or the Peter Thiels, these other people in the world that we point to as doing horrible things, that this would allow them the sort of COVID of the dark of, oh, AI did this, AI launched that missile, AI did this, and they can blame it on some rogue AI incident. while there might be actually human beings and human thinking and human programming pulling the levers behind the scenes. So, there's those concerns. And then, what you point to about the headsets and others just putting AI in everything, absolutely. I mean, the military is embracing AI. Right now, we are not at the point, at least in the United States, where autonomous drones being AI-powered drones where there's not even a human being on a joystick and sitting somewhere in the desert just completely acting on Here, your mission is to find these missile sites, but they actually decide when to press the button. The AI does itself. That doesn't exist yet in the United States, but it's absolutely happening in the Ukraine war. It's happening in Gaza. Other militaries are embracing it. And there is, like with the AI thing, there's a push for some sort of global treaty or agreement about autonomous drones, autonomous weapons, sometimes called killer robots, because That is another real concern, you know, that the AI might be used in a different way. So it might not even be that it's gonna turn into, it's gonna break out of its cage and turn into Terminator the second it jumps in, you know, Elon's Optimus Prime robot or whatever. Although I do have those concerns after seeing, I don't know if you've seen these videos, Kim, of like the Robot Olympics happened a couple months ago.

Yes.

Running around. Some of them are falling down and they're silly. But then there's also the most recent was a robot UFC match. And these are like steel machines, like beating the crap out of each other. There was another one I saw a week ago where the first human to fight a, you know, a robot. Now, the robot's being controlled by a human at the moment, but obviously, if things progress, it wouldn't be that hard to imagine somebody putting an AI bot in there and then whatever, doing whatever. So I think there are layers to these concerns, some that might be overblown. I don't think we're gonna be at Terminator tomorrow, but I am definitely still, you know, concerned. The last thing I will say is just the way that on the most basic level, even if none of these other things are legitimate and never come to pass, the most basic level, it will accelerate and expand the ability for governments to conduct surveillance. AI facial recognition powered together, like that is without a doubt happening and going to get worse.

Mm-hmm. Like flock cameras and whatnot. Exactly. How that's using AI to track from camera to camera to camera. Yeah.

Yeah.

And they use, you know, even just with that one, just to bring up like a side note, there was this horrible, horrible story of a woman that ran over a pit bull puppy. Did you actually, did you hear about this one? This woman— I don't know if I saw this one.

I've seen a few.

Yeah, just happened the other day in Santa Ana, California. And this is how they try to get us to go along with flock cameras, right? This is how, these are the types of stories we're going to see to make us go, yeah, you're right, we need these flock cameras. We need to all be surveilled. Basically, this woman, for whatever reason, she had a pit— what the cameras caught was that she She got out of her minivan with a pit bull puppy. So she brought the pit bull with her, put some kibble on the ground to basically lure the puppy to stay in one spot and eat, and then she ran it over several times.

What the heck?

They ended up finding this woman. Yeah. She was a 75-year-old woman in Santa Ana, and they ended up finding her. And now she's being charged with animal cruelty and whatnot. But just in, in that, and they released the video of her doing it, and that You know, seeing that, I just— which I didn't watch it, I couldn't watch it, but what I did see was the comments and people saying, see, this is why we need flock cameras. I mean, that's what they were literally saying. They were saying, this is why we need flock cameras. We need to catch this woman. And then when she was caught, people said, is that because of flock cameras? And I'm, I'm, I don't know if that's why she was caught or not, but it was certainly the narrative. I mean, people were wanting—

Mm-hmm.

The AI surveillance because of these heartstring terrible stories. You know, and that, and that's how they're going to get us. They're going to show us kids that are kidnapped and killed, and they're going to say, if we had flock cameras, we could have stopped this.

Absolutely. They'll use fear. They'll use, you know, they use all the same old tactics, either saying, look at this horrible person, if only we had flock, or they'll share the success stories. This child was, uh, you know, was found, which is great. Like, I mean, missing children are found, whether it's through flock or any, any, you know, form of technology. Great. But I think that's the bigger point, is do we want to shift further into that world where we continue to trade our our privacy, our individual liberties for convenience and for safety. And that's kind of been this age-old tension that's been existing since the beginning of the United States and since forever. Like, do we want more security? Do we want more freedom? And what is the right balance for that?

Yeah. And I mean, just even the other day, Netanyahu was bragging about how they can get onto your phones. You know, they could get into the phones and You know, he's like basically bragging about the pager, the pager operation and how they basically have access to anything and everything. And, you know, and it's so scary to think like they, they can, they can hack in, you know, especially if these AI companies and these chatbots we're all using, you know, if they've been infected with a Pegasus-style software and they're getting in there and they're, they're able to spy on all of us and they're able to like actually get into our computer software systems, they can plant all kinds of stuff like child porn. onto your computer and then say, see, you're a big pedo. And then that's how they take you down when they want to silence you. And I think that's what they will do to many. I mean, I— and then that's scary because there are actual pedophiles out there who do really terrible things to children. But it's gonna get to the point where we won't know who's who. You know, we won't know, is this just a psyop and this person's not really a pedophile, or is this person and we need to actually prosecute them and they're doing really awful things, or are they just being targeted and they're being taken down by the powers that be that don't want them. And so they planted this stuff on their computer. I mean, that's— and because they're going to do that, it is going to basically undermine the investigations into real incidences of people who harm children. But I think that's where we're headed, unfortunately, because I just don't trust any of it.

I think we're unfortunately there, Kim.

I think you might be right on that. Because I do think that there have been people accused, and people automatically right now, the knee-jerk reaction is to say, yeah, right, you know, this person was a dissident and this is what you've done to them. So yeah, it's a scary world. I mean, I don't— you know, but then I think of Y2K, you know, and everybody was freaking out.

Hopefully this is just that.

Right, you know, it's like, oh my gosh, the world's gonna end the year 2000, everything's ending. And, and, you know, the chips were only set to go to 1999 or whatever it was, and now everything's going to go down. And then nothing happened. I mean, I don't think— I don't remember there being anything that happened at Y2K. I don't think a single glitch occurred. Like, it was just, oh, they just— I don't know that. I don't remember. But I don't remember there being any issue. So I kind of wonder, you know, same thing with the advent of the internet. People thought, oh no, the internet is here. And so all these terrible things are going to happen. And we've just found the internet to be quite a useful tool. So, I just wonder if AI is going to go in that direction rather than it being this doom and gloom, oh, no, we all have to watch our backs, it's coming for our throats.

Yeah, I hope for the best. I hope for the best because I definitely, as I shared today, I think there's legitimate reasons to be concerned and we should also be skeptical because of the financial motivations, because of the other reasons people might benefit by keeping the public afraid again.

Yeah. Okay. Do you stay off of AI entirely?

I don't really use it. I mean, I use occasionally for like, like what you mentioned, like in the sense of a search engine. The most extensive I used it, like when I was making The Pyramid of Power. I'll use AI adjacent tools, I guess you could say, occasionally fixing audio.

Excuse me.

Sometimes I've used it to transcribe documents, like I found 100-year-old letters that I was using for my research, and I could spend the whole day transcribing it and winking, you know, blinking, trying to like make it out, or I could feed it to an AI tool that mostly gets it right and then I correct it with my human hand. But yeah, I'm not completely embracing it, but you know, I don't fault people who do. But I definitely— I will say I definitely don't use it to write my articles, to write my emails, to write my social media posts.

Yeah, I mean, I think it's like a calculator. I mean, that's the way I use it. It's, it's similar to a calculator in that it's a tool That if you pump in the information, it might give you the right answer. Calculator is definitely more consistent and accurate with the answers. But like you said, like if you feed it a letter and you're like, I just need you to transcribe this letter, it'll, it'll do that task. And so it's kind of like using, you know, a calculator or something in that way. But yeah, otherwise pretty scary. Well, excellent work, Derek. Thank you for being here with us and explaining this. I think it's really important information for people to know. Because we are hearing these very scary stories of AI escaped, and that immediately does make us think, oh no, it's smarter than us and there's nothing we can do to contain it, and it's just going to rule the world when we give it enough power. But, um, it's great you've exposed— well, actually it was this one Israeli company that just left the door open and then was like, whoa, we don't know how it escaped. Oh, whoops, we left the door open. That's how. So left the back door open and all the dogs ran out. So, yeah, thank you for having me.

I appreciate it.

Of course. The, and the name of the article once again is Irregular: The Israeli Company with Intelligence Ties at the Center of the Rogue AI Incidents. That is on thelastamericavagabond.com, but we do have the link down below to that. We also have the link down below to the Pyramid of Power. Derek has an excellent series that I do encourage everybody to watch, the various different episodes of Pyramid of Power. Really highly educational, a lot of different— what would you say were the most popular ones in your series, Derek?

I think that The Technocratic State, I think, which is chapter 12, is really popular because obviously more people are questioning technocracy the last few years. And then behind that, I would say, are probably the conclusion, because the conclusion, which is 2 parts, is me attempting to answer the question, well, okay, now that we've described the whole structure of this thing, who's at the top or what is at the top? And so that that got a lot of people curious, I think, and also made a bunch of people think I'm a shill because I didn't point to their particular pet group.

That's controlling the world. Derek Brose, also the author of his memoir, A Man of My Word, which you can also grab. We do have the link down below to that as well. Derek, thank you again for being here. Really appreciate it.

Thank you, Kim.