Former Anthropic Researcher Resigns, Warns AI Could 'Kill Us All' by 2030

Jacob Coxon, a pre-training researcher with experience at the world's leading AI laboratories, resigned from Anthropic on September 8, 2026. His departure was accompanied by a public warning that the pursuit of superintelligence has evolved into a dangerous gamble with human survival.

Upon leaving the company, Coxon took to X, formerly Twitter, to assert that frontier AI companies are "gambling with our lives." Having spent three years conducting pre-training research across both OpenAI and Anthropic, Coxon claimed that the internal sentiment among developers is far more dire than public corporate messaging suggests. He stated that the people actually building these systems "earnestly believe that it could kill us all by the end of the decade."

The Existential Risk Debate

This resignation injects new urgency into the ongoing "AI Doomsday" debate, which focuses on the potential for Artificial General Intelligence (AGI) to become uncontrollable or misaligned with human values. While many critics view existential risk as a theoretical or philosophical concern, Coxon’s warnings stem from the technical front lines of model development. The debate centers on the fear that a self-improving system could rapidly surpass human intelligence, leaving its creators unable to implement safety guardrails or shut the system down.

Industry Implications

Coxon's exit highlights a growing internal tension within the industry's most prominent labs. Researchers are reportedly caught between the intense competitive pressure to achieve a breakthrough in superintelligence and the perceived existential risks associated with that success. Because Coxon has direct experience at both OpenAI and Anthropic—the two most influential players in the frontier model space—his claims lend significant credibility to the argument that extinction risks are a recognized internal concern rather than an external critique.

What to Watch

As the race toward AGI accelerates, the industry now faces increased scrutiny over whether competitive incentives are overriding safety protocols. The primary question remaining is whether other researchers within these labs will echo Coxon's warnings or if his perspective represents a minority view. Observers will be watching for shifts in safety transparency and regulatory responses as the 2030 timeline cited by Coxon approaches.