Sex, AI, and the Apocalypse

In September 2026, a researcher named Jacob Coxon resigned from Anthropic, an AI safety company, and publicly stated that AI researchers genuinely believe artificial intelligence could kill humanity by the end of the decade—a warning that went viral and was corroborated by colleagues at the lab. The article examines how warnings about existential AI risk originate from a specific online subculture of rationalist thinkers whose personal networks, relationships, and ideological commitments heavily influence major AI safety organizations and policy discussions around AI governance.