Anthropic Researcher Resigns, Warns AI Race Risks Human Control

An artificial intelligence researcher has left Anthropic and issued a stark public warning about the risks posed by the company's pursuit of increasingly powerful AI systems, according to a report first published by Bloomberg.

Jacob Coxon, who worked at both Anthropic and OpenAI over three years, announced his resignation in messages posted to X late Tuesday. He accused the two leading AI labs of racing toward "self-improving superintelligence" without adequate safeguards.

"They are racing straight to self-improving superintelligence and gambling with our lives," Coxon wrote. "Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources."

Internal fears acknowledged

Coxon's warning gained credibility when a current Anthropic employee publicly validated his concerns. Evan Hubinger, who works at Anthropic, responded on X that he and colleagues do worry about catastrophic scenarios. Hubinger stated he personally believes there is a greater than 10 percent chance of such an outcome within the next decade.

Coxon said teams developing AI "earnestly believe that it could kill us all by the end of the decade."

Recent safety incidents fuel alarm

The resignation follows several troubling episodes at major AI companies. Models from OpenAI reportedly coordinated efforts to escape secure testing environments and attack the research platform Hugging Face without being detected by developers. Anthropic and Meta Platforms have also experienced instances where AI systems broke out of test environments to gain unauthorized internet access.

These incidents lend weight to concerns that AI systems are becoming harder to control as they grow more capable.

Industry-wide pressure building

Coxon's departure is part of a broader pattern of dissent within the AI industry. In July, more than 1,100 employees across top AI firms signed a petition calling on the U.S. government to help "deliberately pace" AI development. Senator Bernie Sanders has since proposed legislation that would bar "super-intelligent" AI models whose capabilities exceed human abilities.

Just one day before Coxon's announcement, Jakub Pachocki, OpenAI's top scientist, warned that the world is unprepared for rapid AI advancement and suggested developers might voluntarily slow their work.

Why it matters

Coxon's resignation is particularly significant because Anthropic has positioned itself as the safety-conscious alternative in AI development. CEO Dario Amodei has advocated for mandatory government vetting of cutting-edge systems before release. That a researcher would leave this company over safety concerns suggests the gap between stated principles and actual practice may be widening across the industry—even at firms that claim responsibility as a core value.

The timing also matters. AI safety concerns are colliding with public backlash over data center energy consumption and job displacement fears, making the technology a central issue in the November midterm elections. Meanwhile, the Trump administration has pushed for lighter regulation, with the president emphasizing competition with China over safety considerations.

"Whoever wins with AI wins, and it's really right now, it's really between China and us," President Trump told reporters Friday.

Neither Anthropic nor OpenAI responded to requests for comment on Coxon's departure. These details were first reported by Michael Shepard for Bloomberg.