An Anthropic researcher has resigned from the company, warning that the race to build more powerful AI systems is moving too fast. Jacob Coxon, who previously worked on pretraining research at both Anthropic and OpenAI, said neither company is acting responsibly. In a series of posts on X (formerly Twitter), Coxon argued that we should not underestimate the power of AI. The researcher also claimed that some people working on AI privately fear that the technology could eventually pose an existential threat to humanity.
“These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing,” Coxon wrote.
Also read: After MeitY action, NCPCR summons Meta India chief over alleged illegal ads
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
— Jacob Coxon (@hilbertspaess) September 9, 2026
He also claimed that the people developing advanced AI are aware of the risks. According to the researcher, “people building AI earnestly” believe that AI could potentially “kill us all by the end of the decade.”
Coxon also explained why companies continue developing the technology despite these fears. He said many people at OpenAI have not fully understood the scale of the risks. At Anthropic, the risks are better understood, but the company is still caught in a race with other AI labs, he claimed.
Also read: Apple event 2026: Date, India timings, expected iPhones and how to watch live
According to the researcher, Anthropic believes it must continue because other companies may not develop AI responsibly. Coxon called this approach a gamble. “Accepting this race and entering the ‘endgame’ is a hubristic gamble,” he wrote.
The researcher also called for more cooperation between major AI companies. Coxon said stronger measures may be needed if companies cannot coordinate. One option could be a temporary ban on improving AI model capabilities.