Sex, AI, and the Apocalypse

The people warning about AI's existential risk are deeply intertwined with the organizations shaping AI development and regulation.

On September 8, 2026, researcher Jacob Coxon resigned from Anthropic, a leading AI lab, expressing deep concern that AI could cause human extinction by the end of the decade, a sentiment echoed by Anthropic's head of alignment science. This event amplified existing anxieties about AI safety within a prominent internet subculture known for its focus on rationality and clear thinking. This subculture, which originated over twenty-five years ago, has deeply influenced the AI industry, with its members staffing key AI labs, safety organizations, and philanthropic foundations that define AI safety discourse. The group's history includes unusual origin stories, such as a Harry Potter fan fiction serving as a recruitment tool, and its alumni have connections to political influence and unusual living arrangements. The article traces the origins of these concerns back to Eliezer Yudkowsky, a foundational figure in the transhumanist and rationalist movements, who co-founded the Singularity Institute for Artificial Intelligence (now MIRI) in 2000, driven by a conviction that he alone understood the existential risks posed by advanced AI.

The prominent subculture, characterized by a self-perception of superior clear thinking, has cultivated a distinct set of beliefs and practices that now significantly impact the AI landscape. Its influence extends from the foundational research conducted in AI labs to the policy discussions surrounding AI safety and regulation. The social dynamics within this group, including personal relationships and funding networks, are interwoven with the very institutions entrusted with guiding the future of artificial intelligence.

This history reveals a pattern of individuals and ideas emerging from niche internet communities that gain significant traction within the tech industry and, consequently, influence public perception and policy. The convergence of esoteric beliefs, unconventional social structures, and high-stakes technological development raises questions about the vetting processes and the broader implications of these interconnected networks for the future of AI governance and public trust.