Anthropic researcher who quit over AI safety walked away before his stock vested

Jacob Coxon, a 27-year-old Anthropic researcher, resigned from the company in September due to safety concerns over AI. Coxon, who had worked at OpenAI for three years before joining Anthropic, lasted only four months at his new job. He gave up his entire stake in Anthropic, forfeiting equity that would have vested in two months, in order to resign.

This move came as Anthropic prepares for one of the largest IPOs ever, with a reported $2 trillion valuation. Coxon's departure highlights the growing pressure in the AI industry to prioritize safety. He criticizes both Anthropic and OpenAI for racing towards superintelligence without proper oversight, suggesting that the intense competition could lead to cutting corners.

Coxon's concerns include excessive paranoia about competitors like OpenAI and China, and he argues that safety work is sometimes compromised in the pursuit of progress. Despite Anthropic's genuine commitment to safety, Coxon worries that the relentless pace of development could lead to compromised oversight. In a recent statement, over 1,000 researchers signed a call for governments to coordinate and potentially slow AI development once models start improving on their own. Anthropic has not commented on Coxon's departure.