Sex, AI, and the Apocalypse: How a Rationalist Subculture Shaped AI Safety, Politics, and Controversy

The headline warning and its immediate fallout

Jacob Coxon quit Anthropic on September 8 2026, publicly stating that the people building AI “earnestly believe it could kill us all by the end of the decade.” His resignation note went viral (over 100 million views in a day) and was amplified by politicians, Elon Musk, and mainstream media. Within hours, Anthropic’s alignment lead Evan Hubinger confirmed the lab’s internal estimate of a >10 % chance of human extinction within ten years.

The rationalist subculture that fuels the warning

The most vocal AI‑doom advocates belong to a 25‑year‑old internet subculture that calls itself the rationalist community. Originating in early‑2000s forums such as LessWrong, the movement grew around a core set of ideas:

Heroic responsibility – individuals who understand the risk must act to prevent catastrophe.
The Sequences – a curriculum of bias‑overcoming essays that promise clearer thinking.
Rituals and symbols – events like the Secular Solstice (a yearly “doomsday” ceremony) and fan‑fiction such as Harry Potter and the Methods of Rationality (HPMOR) that embed the community’s worldview.

These cultural technologies bind members together, provide a sense of belonging, and create a credential hierarchy where “rationalist” status signals superior judgment.

Key figures and their influence

Person	Role	Notable Contributions
Eliezer Yudkowsky	Founder of the Singularity Institute (now MIRI) and author of HPMOR	Popularized AI‑extinction risk; created the “heroic responsibility” narrative.
Scott Alexander	Blogger (Slate Star Codex) and interpreter of rationalist ideas	Brought rationalist thinking to a wider tech audience; linked AI risk to mainstream Silicon Valley.
Curtis Yarvin (Mencius Moldbug)	Neoreactionary writer	Introduced anti‑democratic “patchwork” concepts that later resonated with some rationalists.
Raymond Arnold	Organizer of the Secular Solstice	Built the community’s primary ritual, framing AI doom as a religious‑like event.
Peter Thiel	Billionaire investor, philanthropist	Funded MIRI, Yarvin’s Tlon, and anti‑democratic projects; delivered leaked “Antichrist” lectures targeting AI regulators.
Jaan Tallinn	Philanthropist	Financed MIRI, the Future of Humanity Institute, and EA initiatives, linking long‑termist funding to AI risk.
Aella	Sex‑worker‑turned‑researcher, AI‑doom content producer	Runs the “Plz Don’t Kill Us” residency with Grimes; illustrates how sexual networks intersect with AI‑safety circles.

From fan‑fiction to organized action

HPMOR (122 chapters, 661 k words) served as an on‑ramp for many future AI researchers. Its climax, the “Final Exam” challenge, mobilized thousands of readers to solve a puzzle under a deadline, demonstrating the community’s ability to crowd‑source complex tasks. The fan‑fic’s author notes also promoted the Center for Applied Rationality (CFAR), which turned the Sequences into in‑person workshops.

Institutional spillover: AI labs, philanthropy, and policy

MIRI and the Future of Humanity Institute (Bostrom) turned rationalist concerns into academic research, attracting billions in donor money (e.g., Open Philanthropy, Thiel Foundation).
Effective Altruism (EA) merged with rationalism via long‑termism: future‑person welfare outweighs present‑person welfare, making AI‑extinction the top cause for funding.
CFAR sold “debugging” workshops to tech firms (Facebook, Thiel‑backed startups) and to gifted teenagers, blurring lines between personal development and corporate recruitment.
Leverage Research and Black Lotus attempted to “understand the mind” but were accused of psychological abuse and cult‑like control.
Zizi LaSota and her followers (the “Zizians”) turned community doctrine into violent actions, resulting in multiple murders and a high‑profile shootout in Vermont (2025).

Political penetration

The rationalist‑EA pipeline funneled money and ideas into U.S. politics:

Sam Bankman‑Fried (FTX) was an EA product who funneled illicit profits into Anthropic and AI‑risk grants before his 2025 conviction.
Peter Thiel financed anti‑democratic campaigns, Gawker lawsuits, and the “Antichrist” lectures that framed AI regulation as a religious‑style oppression.
JD Vance (U.S. Senator) received $15 M from Thiel’s venture firm and cited Yarvin’s anti‑bureaucracy ideas in policy speeches.
Michael Kratsios (former OSTP chief) and David Sacks (science‑tech council co‑chair) are Thiel‑linked appointees, illustrating the network’s reach into federal science policy.

Conflict of interest in oversight

The nonprofit METR, tasked with evaluating frontier AI systems, disclosed in its 2026 risk report that staff have close personal ties to lab employees, raising doubts about its independence. The same small pool of individuals (ex‑colleagues, housemates, spouses) often serves as both evaluator and subject, undermining external accountability.

Community criticism and self‑reflection

CFAR publicly apologized in 2019 for mishandling abuse allegations against Brent Dill, acknowledging that the organization failed to protect victims despite its expertise in judgment.
Former members regularly post “exit threads” describing psychological harm, coercive relationships, and the difficulty of leaving a network where professional, romantic, and housing ties are intertwined.
Commenters on the original article highlighted concerns about groupthink, guilt‑by‑association, and the need to evaluate AI risk on its merits rather than the personal quirks of its advocates.

Why the Coxon resignation matters

Coxon’s departure illustrated two facts:

Credibility of the warning – Even Anthropic’s own alignment lead confirmed the lab’s high existential‑risk estimate, suggesting the threat is taken seriously inside leading AI firms.
Perception of coordination – Critics (including Elon Musk) framed the viral spread as a coordinated psy‑op, pointing to the community’s dense funding and communication networks as evidence of possible orchestration.

Both interpretations are plausible because the rationalist scene is a highly entangled social graph where the same individuals fund, advise, and promote each other’s projects.

The broader lesson

The rationalist community demonstrates how a niche subculture can evolve into a powerful epistemic elite that shapes technology, philanthropy, and public policy. Its strengths—rigorous reasoning, crowd‑sourced problem solving, and long‑termist motivation—are offset by:

Credentialism that equates “rationalist” with “superior judgment.”
Cult‑like dynamics (rituals, heroic responsibility, psychological debugging) that can suppress dissent and enable abuse.
Concentration of influence that creates conflicts of interest in oversight bodies.

Understanding this network is essential for anyone assessing AI‑safety claims, designing regulatory frameworks, or evaluating the legitimacy of the voices that dominate the AI‑doom narrative.

The analysis above is based solely on publicly available reporting and the author’s own citations. Errors or omissions can be reported to the author’s email address.