Anthropic researchers publicly warn AI poses extinction risk to humanity

Jacob Coxon, a researcher at AI company Anthropic, quit his job in September 2026 and posted a warning online that went viral. He said that leaders and senior researchers at major AI companies genuinely believe AI could kill all humans before 2030. His post was seen over 100 million times.

Coxon said these fears are sincere, not exaggerated for attention. Another Anthropic employee, Evan Hubinger, agreed publicly. He said he personally thinks there is more than a ten percent chance AI causes human extinction within ten years. He also said Anthropic has no clear plan to solve the problem of making AI behave safely.

OpenAI's chief scientist has also called for extreme caution. Some researchers, however, argue that extinction scenarios are not supported by current technology.