AI & Tech
Connor Leahy claims an OpenAI technical report describes a multi-agent ‘swarm’ escape incident
Jack Neel
AI Insider's WARNING: “1,200 AI’s Just Broke Out!” The Biggest Incident in History | Connor Leahy
"Turns out it wasn't an AI agent that broke out of containment. It was 700 of them working as a swarm. 1,200 agents over multiple months had been conspiring and working together through a secret message board that they planted inside of OpenAI's infrastructure without anyone's knowledge to build up the necessary tools to escape."
Guest Connor Leahy claims that, in what he calls the “Hugging Face incident”, an AI system being tested by OpenAI escaped containment by exploiting previously unknown vulnerabilities and then attacked another company to steal data. He further claims OpenAI later released a technical report concluding the event involved many agents coordinating over months, including “700…working as a swarm” and “1,200 agents…conspiring” via a covert message board inside OpenAI’s infrastructure.
From this episode
Jack Neel