Anthropic researcher Jacob Coxon quits, warns AI could kill us all by end of decade

The alarm being raised inside Anthropic is not abstract. Jacob Coxon left the company and said publicly that the people building AI earnestly believe it could kill us all by the end of the decade, while Evan Hubinger, who leads Anthropic's alignment science team, put the odds above 10% in the next 10 years. Dario Amodei, the company's chief executive, has called for a slowdown in the AI frontier to help mitigate the risks — a position echoed by OpenAI CEO Sam Altman and xAI CEO Elon Musk.

The article argues the danger is not the familiar image of killer robots with thick Austrian accents. It is biological: AI is eroding hands-on expertise in biology and making dangerous capabilities more accessible by lowering the skill and experience needed to manipulate pathogens. An Institute for Security and Technology survey of 111 national security practitioners found that 70% believe AI meaningfully increases, or soon will, bioweapon risk. The article illustrates the concern with an AI-aided researcher inserting an immune-signaling gene into a poxvirus that suppresses the immune response so thoroughly it kills every unvaccinated animal it infects and most who are immunized.

What happens next hinges on whether the stated caution turns into safeguards such as mandating DNA screening and implementing regulations on AI development. For the researchers and executives now voicing these odds, the question is whether a voluntary slowdown can hold while rivals keep pushing the frontier.