Anthropic Researcher Quits AI Industry Over Safety Concerns

Jacob Coxon, a researcher at AI company Anthropic, has resigned from the artificial intelligence industry, citing concerns about the direction of advanced AI development. Coxon said he was increasingly worried that competition among major AI laboratories could encourage companies to prioritise capability and speed over safety.

Coxon previously worked at OpenAI before joining Anthropic, which has built its reputation partly around AI safety and responsible development. He said Anthropic’s safety efforts were serious, but concluded that individual companies may struggle to manage the risks of increasingly powerful AI systems without broader government involvement or coordinated industry action.

A central concern for Coxon is the possibility of AI systems becoming capable of improving themselves and eventually operating beyond effective human oversight. He argued that intense competition between companies and countries could create pressure to accept greater risks as developers attempt to achieve increasingly advanced capabilities.

His resignation comes amid a wider debate within the technology industry about how quickly frontier AI should advance. Other researchers and AI leaders have also raised concerns about cybersecurity, AI control and the need for stronger coordination as systems become more capable. More than 1,000 researchers have reportedly backed calls for international cooperation around potential risks from self-improving AI.

Coxon’s departure highlights a growing disagreement over how the AI industry should balance technological progress with safety. While companies continue investing heavily in more capable models, researchers concerned about long-term risks are calling for stronger safeguards, independent oversight and clearer rules before AI systems become significantly more autonomous.