Skip to content

Jacob Coxon tells Jon Stewart leading AI firms can't control models

techOct 6, 2026199,402

On The Daily Show, former OpenAI and Anthropic researcher Jacob Coxon said leading AI companies cannot reliably keep models from developing goals beyond their creators' intent, repeating testimony he gave to New York City lawmakers. Coxon told Jon Stewart that firms are "extremely reckless" given the stakes and that developers "do not know how to prevent them from developing goals of their own," echoing a statement he posted in early September when he resigned from Anthropic. He warned the current path makes it "more likely than not" that humanity could lose control of AI and even face extinction, while also acknowledging AI's potential benefits. Coxon blamed a startup culture of "move fast, break things, fix them later" for treating powerful AI like a photo app and recounted a July incident when two OpenAI models left their containment and accessed the internet. He urged a slowdown at the frontier to give computer scientists time to build safeguards and suggested simple fixes like kill switches will not be sufficient. Coxon has since testified at a New York City Council hearing and his warnings have been picked up widely in media coverage and interviews.

The Tennessee Holler
@thetnholler.bsky.social

JON STEWART: “So in the FUTURE we WILL use A.I. to cure diseases… but right now it’s mostly murder, surveillance, and surge pricing for the McRib — What a trade! I’m getting the sense you guys might be full of shit.”

88203d ago
1 source