Jacob Coxon, a former researcher at OpenAI and Anthropic, resigned in early September 2026. He publicly expressed concerns that leading AI laboratories are engaged in a competitive race to develop self-improving superintelligence, potentially reaching this milestone by 2027.
Coxon, who contributed to the development of the GPT-4o model during his three-year tenure at OpenAI, claims that industry professionals acknowledge the risk of human extinction from advanced AI within this decade. Despite these concerns, he argues that development continues without sufficient oversight.
The situation has drawn attention from other industry figures, including Anthropic's Alignment Science Lead, Evan Hubinger, who estimated the probability of AI-related human extinction within the next ten years at more than 10%.
Following his resignation, Coxon is scheduled to testify at a New York City Council hearing on AI safety on October 5, 2026. Representatives from Anthropic, OpenAI, Google, and Meta are also expected to participate in the session.
