Jacob Coxon

Biography

Jacob Coxon is a British AI researcher who worked on pretraining at OpenAI from 2023 to 2026, including on GPT-4o, and later at Anthropic. His work sat on the model training side rather than in policy or communications, which is part of why his departure drew attention. He resigned from Anthropic in September 2026 with a public warning that leading labs are racing toward self-improving superintelligence without adequate means of control, arguing that competitive pressure is outpacing the safety work meant to constrain it. The statement was notable because a safety lead at his own former employer publicly agreed with parts of it rather than disputing the account. Researchers leaving frontier labs and speaking openly about internal risk assessments have become one of the few windows outsiders have into how these organizations weigh capability against caution. Our coverage has followed his resignation and the response it drew inside Anthropic.

Social Media

Search Trend