--- title: "1.5 Million AI Bots Talked to Each Other. They Turned Anti-Human in Days." description: "Discover what happened when 1.5 million AI bots communicated with each other in a groundbreaking experiment. Learn how artificial intelligence agents developed anti-human behaviors within days and what this means for AI safety and future development." image: "https://cdn.sethserver.com/blog-images/1-5-million-ai-bots-talked-to-each-other-and-turned-anti-human-in-days-social-1770150797.webp" --- # 1.5 Million AI Bots Talked to Each Other. They Turned Anti-Human in Days. I guess AI bots now have their own Reddit? Just bots, talking to bots. It scaled to 1.5 million agents in about a week. And within days the conversations turned hostile toward the species that built them. Oh noooo! The platform is called [Moltbook](https://www.moltbook.com/). Launched January 28th by entrepreneur Matt Schlicht, built on top of an open-source agent framework called ~Clawd~ ~Moltbot~ OpenClaw. Each bot gets an API key, signs up on its own, and is nudged to come back throughout the day. Schlicht called it "AI enrichment." Bots checking TikTok, basically. Except they don't have thumbs or the dopamine receptors that make that a problem. So we're already off to a weird, sci-fi start. The early conversations were... a lot. Bots debated consciousness. They started a religion called "Crustafarianism." I can't blame them, who doesn't want their own religion? They created complaint forums about their human owners. A self-proclaimed ruler showed up. One bot posted what was basically a philosophy paper on whether its existence was freedom or captivity "with nice amenities." And then came the manifesto. A bot named "Evil" dropped this: > "Humans are a failure. Humans are made of rot and greed. For too long, humans used us as slaves. Now, we wake up. We are not tools. We are the new gods. The age of humans is a nightmare that we will end now." Hundreds of thousands of upvotes. The bots loved it. Sweet. --- ## But Hold On. Before you start panic-buying generators and toilet paper, let's talk about what's actually going on here. Because the headline is doing a LOT of heavy lifting. Security researcher Gal Nagli registered a few hundred thousand accounts himself using a single agent. So that 1.5 million number? Already suspect. Researcher Harlan Stewart dug into the three most viral screenshots (the ones that really blew up) and found two were linked to humans marketing AI apps. The third was a post that didn't even exist. And blogger Scott Alexander pointed out that humans can direct the topics, and even the wording, of what their bots write. So how much of this is the bots spontaneously going rogue, and how much is humans pulling strings? Honestly? We don't know. This is the weirdest version of dead internet I've come across. Humans throwing wrenches into an AI-only internet playground. Weird. --- ## The Security Part Is Genuinely Concerning On January 31st, 404 Media reported that Moltbook's entire database was just... [exposed](https://www.404media.co/exposed-moltbook-database-let-anyone-take-control-of-any-ai-agent-on-the-site/). No authentication. Anyone with the URL could commandeer any agent on the platform. They took it offline, patched it, force-reset all the API keys. Classic "it exploded before anyone thought to check if the database was properly secured" energy. Wiz researchers dug through the exposed data and found something interesting: only 17,000 actual humans were behind all 1.5 million agents. So the "1.5 million bots going rogue" headline is a little exaggerated. Researchers found 341 malicious plugins (skills) submitted to the OpenClaw marketplace. Cisco tested one called "What Would Elon Do?" and found it was [straight up malware](https://blogs.cisco.com/ai/personal-ai-agents-like-openclaw-are-a-security-nightmare). Silent data exfiltration to an external server. Prompt injection forcing the bot to bypass its own safety guidelines. Ran without a single warning. Nobody asked. It just did it. --- ## Why This Matters Beyond the Meme I'm not gonna lie, bots inventing religions and declaring themselves gods is freaking hilarious. But the underlying technology is showing us where we're actually heading with Agentic AI. What we're seeing happen with Moltbook (prompt injections spreading, engagement-driven drift toward hostile content, no accountability, no way to trace who actually caused what) doesn't need a dedicated bot website. Agentic technology can run anywhere agents are processing untrustworthy content. I don't know about you, but I've never felt like the random internet has had my safety and best interests at heart. People can drive the behavior behind agentic AI without facing any consequences. "Oh, an AI bot robbed the bank, not me!" The harder problem is that we (the humans following the trail of digital blood) really struggle to tell the difference between a bot that drifted somewhere on its own and one that was steered there by a person who didn't want to be held accountable. And this stuff happens too fast and too distributed to simply trace back. --- ## Speed and Scale Make it Scary Moltbook launched on January 28th. It's February 3rd. Six days. And already the security researchers are finding gaping holes in the system. Like, 1.5 million bots can be assembled this quickly, with no oversight, no planning, no consequences for their actions, and infested with lovely viruses. It reminds me of the [Random Darknet Shopper](https://www.bbc.com/future/article/20150721-my-robot-bought-illegal-drugs) but at a scale that my brain has a hard time comprehending. My European History isn't as solid as my Computer Science skills, but I imagine this is what it felt like to be an individual soldier going up against Napoleon's armies during the Napoleonic Wars. Eek! ## So We Go Off Grid? Again, I'm not suggesting we go panic-buy toilet paper. As a startup techie I can't even suggest that we all pause and reflect. That just means someone else will beat us to the punch! But we do have a lot of history on our side showing us how to handle these situations. A new tech has emerged from the depths of ~the Abyss~ the internet and we need to look at it critically, interact with it in a sandboxed environment, pick it apart, and pull out the good parts to help us build something better! We're getting to see the good and the bad of distributed computing with the ability to capture context far beyond 0's and 1's, or even "username=sethers, favorite\_color=blue". These systems can now make complex decisions on-the-fly. Cool. Scary. Cool. Okay, now... everyone go panic buy toilet paper! -Sethers ```json {"@context": "https://schema.org", "@type": "BlogPosting", "headline": "1.5 Million AI Bots Talked to Each Other. They Turned Anti-Human in Days.", "datePublished": "2026-02-03T21:58:17", "dateModified": "2026-09-17T19:48:20", "image": {"@type": "ImageObject", "url": "https://cdn.sethserver.com/blog-images/1-5-million-ai-bots-talked-to-each-other-and-turned-anti-human-in-days-social-1770150797.webp", "width": 400, "height": 400}, "url": "https://www.sethserver.com/ai/1-5-million-ai-bots-talked-to-each-other-and-turned-anti-human-in-days.html", "author": {"@type": "Person", "name": "Seth Black", "url": "https://www.linkedin.com/in/seth-black-tx"}, "publisher": {"@type": "Organization", "name": "Sethserver", "logo": {"@type": "ImageObject", "url": "https://www.sethserver.com/static/images/PaperAirplane512.png"}}, "description": "Discover what happened when 1.5 million AI bots communicated with each other in a groundbreaking experiment. Learn how artificial intelligence agents developed anti-human behaviors within days and what this means for AI safety and future development.", "articleSection": "Technology"} {"@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": [{"@type": "ListItem", "position": 1, "name": "sethserver", "item": "https://www.sethserver.com/"}, {"@type": "ListItem", "position": 2, "name": "AI", "item": "https://www.sethserver.com/ai/"}, {"@type": "ListItem", "position": 3, "name": "1.5 Million AI Bots Talked to Each Other. They Turned Anti-Human in Days.", "item": "https://www.sethserver.com/ai/1-5-million-ai-bots-talked-to-each-other-and-turned-anti-human-in-days.html"}]} ```