Former Anthropic Researcher Warns AI Could End Humanity This Decade

A former researcher at Anthropic has resigned with a stark warning: the people building advanced artificial intelligence systems genuinely believe their work could lead to human extinction within the next six years.

Jacob Coxon, who recently left the AI safety company, issued his warning in a public resignation announcement that challenges the industry's breakneck pace. "The people building AI earnestly believe that it could kill us all by the end of the decade," Coxon wrote. "No other human activity poses this level of danger."

The competitive pressure driving risk

Coxon's central concern centers on what he describes as a "race dynamic" among major AI companies. In an interview with NPR's Scott Detrow, he explained his worry that competition is pushing labs to develop superintelligent agents faster than safety measures can keep pace.

The researcher is calling on Anthropic and its competitors to deliberately slow their development timelines. His resignation represents a growing tension within the AI industry between those pushing for rapid capability advances and those advocating for more cautious approaches.

Why it matters

Coxon's warning carries particular weight because it comes from inside one of the companies most focused on AI safety. Anthropic was founded by former OpenAI executives specifically to build AI systems more carefully. When safety-focused researchers at safety-focused companies sound alarms about existential risk, it suggests the concerns extend beyond outside critics to those with direct knowledge of frontier AI development. The acknowledgment that builders themselves believe their work poses civilization-level threats raises questions about whether current governance structures are adequate.

Inside perspective on existential concerns

What makes Coxon's statement notable is his claim about the beliefs held within AI labs themselves. Rather than dismissing extinction scenarios as science fiction or external fear-mongering, he asserts that developers working on these systems take the possibility seriously.

This internal acknowledgment of catastrophic risk, combined with continued rapid development, forms the core of Coxon's critique. The implication is that commercial and competitive pressures may be overriding safety considerations even when the people involved recognize the stakes.

The details of Coxon's concerns and his call for industry changes were first reported by NPR.