AI Insider Quits, Issues Dire Warning About Humanity’s End

Jacob Coxon, a researcher who worked at both OpenAI and Anthropic, has left Anthropic with an unusually stark warning: some of the people building the world’s most capable AI systems genuinely believe the technology could threaten humanity’s survival.

Coxon announced his resignation in a post that went massively viral. Axios reported that it surpassed 115 million views. He told the outlet he left roughly two months before his Anthropic equity would have vested, giving his departure a significant personal financial cost.

His warning was direct. Coxon wrote that people building AI “earnestly believe that it could kill us all by the end of the decade.” Anthropic alignment-science lead Evan Hubinger publicly agreed with the basic concern and said he personally assigns a greater-than-10% probability to AI killing all humans within the next decade.

Those are forecasts, not established outcomes. Other technology leaders reject the doomsday framing; Nvidia CEO Jensen Huang, for example, has publicly dismissed human-extinction claims as exaggerated. The significance of Coxon’s resignation is not that it proves AI will end humanity, but that researchers with direct access to frontier systems are willing to leave valuable positions because they believe the risk is serious enough to outweigh the rewards of staying.

That argument has now escaped the small world of AI-safety research and entered mainstream politics. The public does not need to accept Coxon’s probability estimate to recognize that the people closest to frontier AI are having a much more alarming conversation than most consumers realize.