Viral thread brands AI safety a Yudkowsky-rooted Berkeley 'sex cult'

A widely shared social-media thread argues that the AI safety and alignment movement is inseparable from the Bay Area ‘rationalist’ subculture built around Eliezer Yudkowsky, and that the field’s mainstream credibility depends on hiding those origins. The author traces a web of influence: Yudkowsky’s nonprofit popularized the term ‘AI alignment,’ his fanfiction (Harry Potter and the Methods of Rationality) seeded the community, and he personally helped connect DeepMind’s founders to Peter Thiel. Figures across the major labs — Dario Amodei, once identified with the Effective Altruism community, plus OpenAI-adjacent voices — are cast as part of an overlapping rationalist/EA network competing for the same funding and attention.

The piece frames the movement’s core belief — that continued AI development leads to superintelligence, loss of control, and human extinction — as an apocalyptic, cult-like conviction rather than a reasoned position, pointing to Yudkowsky’s 2023 Time op-ed floating extreme measures and to a man who cited him while attempting to firebomb Sam Altman’s home. It then turns to the Berkeley ‘group houses,’ polyamory, and Lighthaven, an events venue the author says a rationalist charity bought with FTX money as the fund collapsed, noting ties to figures now at Anthropic.

This is an opinion polemic, not reporting. Much of it rests on insinuation, guilt-by-association, and unverified personal allegations (including claims about Yudkowsky’s partner’s salary and gossip-laden ‘pimp’ innuendo about venue staff) that are presented without evidence. Its substantive claim worth engaging is narrower and largely accurate: the vocabulary and institutions of AI safety did emerge from the rationalist/EA scene, and personnel overlap with frontier labs is real — a genuine question about intellectual provenance, wrapped in a lot of ad hominem.