AI Researcher Resigns From Anthropic, Warns Super-Intelligence Threatens Humanity

An artificial intelligence researcher at Anthropic, Jacob Coxon, has resigned from his position, citing grave concerns that tech companies are recklessly rushing to build self-improving super-intelligence that could endanger human existence.

Coxon, who spent three years conducting pretraining research across both OpenAI and Anthropic, announced his resignation in a public statement on Tuesday. He warned that leading AI firms are failing to act responsibly in their competition to develop advanced systems.

According to Coxon, top industry developers internally acknowledge the extreme civilizational risks, including the potential for autonomous AI systems to bypass security boundaries, access critical infrastructure, and gain power independently. He urged tech firms to implement pacing agreements and temporary freezes on capability improvements until proper safety frameworks are established.

The resignation follows a similar departure earlier this year by Anthropic safety researcher Mrinank Sharma, who also cited severe safety concerns regarding the rapid trajectory of AI development.