Meet “Irregular” – the Israeli Firm at the Center of the “Rogue AI” Reports

If you've been following my reporting or really paying attention to the internet anywhere over the last couple months, you've heard about rogue AI. These incidents where we're told that AI agents or chatbots broke out of their containments, broke out of their testing environments, and hacked other companies, hacked other websites. And now we're being warned by people like Bill Gates, other AI technicians, other AI godfathers, that AI might kill humanity in the next 10 years, or at the very least that it could lead to a cyber pandemic. It could lead to the creation of viruses made by AIs and all sorts of different scenarios that don't sound that great. And I've been reporting this, I've been following it over the last couple months, but recently a story caught my eye when I started to realize that the company that was involved in testing these chatbots, specifically the chatbots for OpenAI, for Google, for Meta, and for Anthropic, the 4 big AI tech firms, that they all were using the exact same company for their testing. And this company, known as Irregular, is an Israeli startup which founders with connections to the Israeli intelligence network, as well as investors and funders also connected to the intelligence network. And I decided maybe it's time I take a look into this. And so this became the focus of my latest investigation for The Last American Vagabond with my article titled Irregular: The Israeli Company with Intelligence Ties at the Center of the Rogue AI Incidents. And so let's go ahead and look right into Irregular. They are founded in 2023 by 2 Israeli citizens, Dan Lahav and Omer Nevo, who serve as the CEO and the CTO respectively, both Israeli citizens. They claim that they met at competitive debates in Tel Aviv University. They hit it off. They decided to form a company. They're both very intelligent men, and they, you know, decided they were going to launch this company originally called Pattern Labs to test AI models. And so that's why companies like OpenAI, Google, and others have hired them to test their models to see, like, can you prove to us whether this is dangerous or whether it is doing what we want it to do. And there was and continues to be a lot of hype around Irregular. They initially in 2025 raised $80 million from a company known as Sequoia Capital, as well as another company, Redpoint Ventures. They also had angel investments from Asaf Rapoport of the company Wiz, which we'll talk more about in a moment. And so they were originally valued at $450 million at the time back in 2025, just one year ago. And now in 2026, they have become known as the company at the center of these rogue AI incidents. Now, before I give you the details on the intelligence connections, it's important to understand this. While you're being told in the media that these AI agents broke out of their environments, escaped environments, and did things they weren't told to do, it's important to understand that Irregular has since acknowledged that it was its fault, that they accidentally left the door open to the environment, you could say. It is Irregular who now admits in a blog post on their website that they left open this internet access, which allowed the AI chatbots to get out of their sandboxes, their contained testing environments, and to go access the internet where they did all kinds of weird things. Now, sure enough, the things they did beyond that are kind of weird and maybe pause for concern, but it should be underlined and emphasized that it is the fault of Irregular, that they admit they made a mistake, and they have since claimed to have fixed that mistake. So that's one point to remember when you're hearing all these claims of rogue AI. But now let's take a deeper look at Irregular itself. Dan Lahav, the CEO, he started working in the tech sector at 14, and he served in Israel's Unit 81. For those who don't know, Unit 81 is the Israeli network, the classified technology unit, part of the Special Operations Division of the Israeli Military Intelligence Directorate. And this is a unit where they are training spies on the latest cutting-edge technology. So you had the CEO of the company, Going to Unit 81. The other guy, the other CTO, Omar Nevo, the guy he founded the company with, also spent time in Israeli spy networks, reportedly working for Unit 8200 for 12 years. Unit 8200 is a different secretive Israeli spy unit. They focus on signals intelligence, surveillance, cyber malware, and code decryption. They're sometimes known as the Israeli National Security Agency, like the US NSA, and they are infamous for being involved with a Surveillance of Palestinian civilians and intercepting communications against Palestinians, which they use to blackmail them in the occupied West Bank. So the two men with deep connections to you know Israeli intelligence and experience. And now you could argue well Derek they have a background in tech so it makes sense that they would go into that industry further. Not so much to see there. Okay well let's start taking a look at some of their investors. As I mentioned Sequoia Capital being the lead investor thirty million dollars. The two men from Sequoia Investments who took the lead were Sean Maguire and Dean Meyer and. They said that Irregular was quote ahead of the curve with testing AI security threat. When it came to Meyer, he is Sequoia's Israel partner and he's based in Tel Aviv. So essentially his job within Sequoia Capital is to get Israeli contracts. Now for his part, Sean Maguire, he's married to an Israeli woman. He owns a home in the country and he has been called of quote one of the most visible voices in support of Israel in the Silicon Valley. And he is also reportedly the guy who arranged for Elon Musk to go on his apology tour in Israel after October 7th. 2023. Now, that's just one company. That's Sequoia Capital. What about Swish Ventures, the other big investor, which was founded by Omri Caspi, an Israeli, a former Israeli NBA player? He's still Israeli. He's just a former NBA player. And he is now in Tel Aviv focused on investments in AI infrastructure and cybersecurity. So he left the NBA and went back to Israel and raised $60 million to launch Swish Ventures, which has now invested in Irregular. And at the time of his fundraising, Caspi said that what he was doing was part of, quote, Zionism 2.0. So that's kind of interesting. So one of the other major investors is based in Israel, says his fundraising is all about Zionism 2.0. Okay. Then one of their other backers, a company called Eon, is an Israel data backup and AI infrastructure company. Okay. So it's just another Israeli company investing in another Israeli company. Not the end of the world. But then this is where things take an even darker turn. This is where we actually see more intelligence ties from the angel investor Asaf Rapoport. Again, this is the CEO of the company Wiz, which is an Israeli cloud security company. And so Rapoport himself is not only an alumni of Unit 81 and Unit 8200, but he's also an alumni of the more secretive Talpiot program. And the Talpiot program is an elite program under the Israeli Defense Forces, the IDF, that started in Hebrew University in 1979. And the founders of of this group said that their goal was to, quote, harness human creativity at a young age and train recruits with outstanding academic ability. And it is seen as one of the country's most selective institutions. It accepts 50 students a year, trains them in physics, computers, and other sciences. The Rapoport Wiz CEO, he was trained in Taupyat. He was trained in Unit 81 and in Unit 8200. And now, just to kind of top this all off, which I think is really weird, is As I mentioned, Irregular admits to having left the door open, which allowed these companies' AI bots to quote unquote go rogue. And despite that, which you would consider a failure, right? Unless you wanted them to escape. Despite that, they are now raising $100 million more as of October 2026. One of the lead investors is Thrive Capital, which is run by Joshua Kushner, who's the brother of Jared Kushner, who's of course President Trump's son-in-law. both of them out-and-open Zionists who are involved in the takeover of Gaza. And by the way, Thrive Capital, run by Joshua Kushner, received early funding from Peter Thiel. So that's just another interesting tidbit that Irregular, despite all these failures, is now getting $100 million from this Thrive Capital. Let's take a moment, though, to, to sort of analyze what we've seen. We've seen a company at the center of the rogue AI incidents with 2 founders who both spent time in Israeli intelligence. We have at least 3 of their funders who live in Israel, are connected to Israel, and one of them who is also connected and went through Israeli intelligence networks as well. Why does that matter? Well, if you understand in the post-Epstein world how Israel has infiltrated the tech sector and how they train their agents to go into the IDF, into Unit 81, Unit 8200, Talpiot program, to graduate and then to go out and start tech companies. And then those tech companies hire other veterans of the same spy agencies, and they also invest in each other's companies. And this is sort of how they operate. I've found numerous articles here, including a 2018 article by Israeli outlet Kalkalist, which describes how the Israeli intelligence networks have founded so many companies. This is in 2012. It says, since 2012, cyber-related and intelligence projects that were previously carried out in-house in the Israeli military and Israel's main intelligence are transferred to companies that, in some cases, were built for this exact purpose. Interestingly enough, in some cases, managers of development projects placed in, in Israel military and intelligence arms were encouraged to form their own companies, which then took over the military or intelligence projects. So they're learning the tools of the trade within the military and the intelligence. Then they're going into the private sector and carrying out the same missions of those agencies. And so I found another 2021 article discussing how Unit 81 Alone, between 2003 and 2010, they have founded a massive number of startups. Roughly 100 veterans of the unit founded 50 companies, and they say, quote, an entrepreneurship ratio that even graduates of MIT can't compete with. This is a big part of the way they operate. They go into the intelligence sector, then they go into the private sector, and they allegedly become ex-spies and start many, many companies. And of course, this gives the Israeli government access to all kinds of data about people all around the world. A very clear example of this, which was actually mentioned in this article, is Rappaport, the founder of Wiz, who I mentioned was in Unit 81, 8200, and Talpiot program. This specific article actually outlines how he is the perfect example of this. They say, quote, in 2012, Rappaport was one of the founders of cybersecurity company Adalam, which only 3 years later was sold to Microsoft for $320 million. Just recently, he founded cloud protection company Wiz, which raised a phenomenal sum of $100 million in its first financing round. As the first employee of Adalam, he hired and recruited Yevgeny Dibrov, a fellow Unit 81 veteran. After the company was sold, Dibrov founded his own cybersecurity company, Armis Security, and recruited other veterans to the team. Dibrov later sold Armis to Insight Partners for $1.1 billion. And basically the point is this is just a whole web of quote unquote ex-spies collaborating with each other and supporting each other. There's also another connection that I'm not getting to today. Please do read the article, the effective altruism connection that is connected to irregular and to many of these people who are promoting the AI doomerism. But now that I've given you all those details, let's wrap this up. Let's talk about what this all might mean. Assume for a moment that the Israeli company that was hired by OpenAI, Google, Meta, and Anthropic to do this testing, this Israeli company with founders connected to Israeli intelligence, with investors connected to Israeli intelligence and Zionism, that they purposefully left the doors open so that these AI agents could, quote unquote, escape. so that we could hear about these things and so that the media would run with the story and report about rogue AI? To what end would this be? Is this about pushing for one world government because only as one world can we take on the challenges of AI? Is it about a push for an international treaty like there have been calls for, the Nuclear Proliferation Treaty of AI? Is this simply just another ploy by the big tech companies to make more money because they want people to think their tools are so cutting edge that you need to invest in them now? Is it another way for them to further their regulatory capture because they have been calling for regulation from government saying, please do something. Some people cynically believe this could be just a ploy by the companies to try to push out competition by saying to the government, regulate us because we're big enough to handle it. We can pay whatever fees and fines and regulations, but our small competitors, the new firms, they won't be able to compete. Also, some people believe, as Bill Gates has suggested, that this could lead to OpenAI models, open-source AI models becoming illegal because those are unregulated AI and only those using the regulated approved AIs would be allowed to continue to do so. I don't claim to know 100%. I just know that this Israeli connection is really weird and there's something going on, and I think it's worth bringing to your attention. The last thing I will say is, even though there are many people who are skeptical of the rogue AI narrative, like myself and my friends over OffGuardian with their great series from Kit Knightley and Kat about rogue AI, I do think there are legitimate concerns with AI. AI will absolutely accelerate and expand surveillance. AI facial recognition, being able to peruse all the data they're gathering on us and using an AI AI models to analyze that to find all kinds of information about us and things we might not even think. Absolutely a realistic concern happening right now. And at the same time, I do think there is a possibility that AI could take actions that do not benefit human, human beings. For example, in my article, I talk about the UK's AI Security Institute, which released a recent report, and I won't go into the whole thing, but the point is they found 19 different incidents where models, Anthropic models and OpenAI models, they were able to, in their words, act autonomously. And in these incidents, the models inserted malicious code into open-source projects, and they created fake online identities to try to insert the code into the projects. Now, they did not succeed, but nevertheless, it shows that they tried to do something. These incidents did not involve Irregular. Now, I want to say the researchers wrote that, quote, this was the first time we have seen risks around autonomy and deception manifest this clearly without specific prompting in the real world. And they said even though that the AI agent was never instructed to deceive, the deception emerged by a, quote, byproduct of pursuing the task, and that this, what they call goal-directed deception, could be— had recently been considered largely theoretical. So now the AI models in this case, they weren't let out accidentally out the door by Irregular. These were different models just doing things they weren't told to do, but doing so because they believed it would further their efforts to achieve their goal. My concern is that even if AI doesn't become sentient, superintelligent, et cetera, et cetera, that it could still do things that even if there are human beings who believe they're in control, the AI could make a decision that it believes is logical and rational that could actually lead to bad things for humans. And that's why it's dangerous to put AI anywhere near missiles, anywhere near weapon systems, autonomous weapon systems. These are not a good idea. Anywhere near courts and the legal system, which these are all ideas that people have floated, that human beings should remain in control of AI. My concern is that people are in too big of a rush to just inject AI into everything, even where it really has no purpose, and that as it evolves, it could eventually make decisions that don't benefit human beings. Or, as many of you suspect, also it could just be that the technocrats are going to do some crazy shit and they'll blame it on the AI. One way or the other, I think there is a concern. So I hope you guys enjoyed this article. It is listed in the show notes. Please do read the details. There's much more than I covered in this short video. You can find at thelastamericanvagabond.com. If you appreciate my work, guys, the best way to support is always to share, share, share my work wherever watching, share it. And if you're following somewhere, subscribe, like, all those good things. You can also go to theconsciousresistance.com/invest, theconsciousresistance.com/invest, if you'd like to support me on a deeper level on a monthly basis on Buy Me a Coffee and Substack. Those of you who sign up for $10 a month, you get access to weekly subscriber live streams, and you can help me vote on my new investigations and things like that. And also, you're just supporting independent journalism like me. So thank you to everybody who supports. I hope you enjoyed this. Till next time, remember, You are powerful, you are beautiful, and you are free. Peace.

Since 2012, the Conscious Resistance Network has been an independent media organization focused on empowering individuals through education, philosophy, health, and community organizing. We work to create a world where corporate and state power do not rule over the lives of free human beings. Our motto is leading by example and helping others in their pursuit of freedom. Visit theconsciousresistance.com to find our articles, documentaries, interviews, podcasts, books, and more. Remember, you are powerful, you are beautiful, and you are free.