Silicon Sickos: Anthropic Employee, Other AI Experts Run 'Consensual Non-Consent' Sex Club
Context:
An investigation reveals a Bay Area sex party group named Red Means No, founded in part by AI industry workers, including Anthropic staff, operates with a consent-based but boundary-pushing framework and has been linked to a sexual assault claim. The group hosts menacingly explicit events described as allowing ‘default consent’ to physical acts, with safe words to stop, while more extreme acts require negotiation. The roster features several AI figures connected to Anthropic, MIRI, and Eleos AI, prompting concerns about industry pressure on women to participate for career advantage. Critics argue the culture surrounding Silicon Valley AI labs fosters networking through sexuality, potentially signaling problematic character traits to assess safety experts. The case underscores tensions around ethics, workplace culture, and the social dynamics shaping AI research communities, with ongoing scrutiny and calls for accountability.
Dive Deeper:
Red Means No is a California LLC formed in 2023 and operates from the San Francisco Bay Area, hosting a series of sadomasochistic parties described as spaces where attendees may ignore another’s refusals, with a safety word ('red') to stop encounters.
The regime permits more extreme acts like light electricity or moderate beatings only with prior negotiation, while prohibits body-shaming and race-shaming, and has at least one reported sexual assault tied to its events.
Four co-founders are identified: Paul Crowley (Anthropic cybersecurity researcher), Aaron Silverbrook (former MIRI director of operations), Rosie Campbell (Eleos AI managing director), and Aella (an OnlyFans writer and organizer cited as a creator of the parties).
Aella has publicly described starting the parties after a personal experience at a vanilla orgy, and is connected romantically to Nate Soares, president of MIRI, linking the group to notable AI safety figures.
Campbell previously worked on OpenAI’s safety team and later evaluated model welfare for Anthropic’s Claude 4; she attended AI safety meet-ups organized by Aella, though later filings did not list her as a Red Means No manager.
The Washington Free Beacon report notes a disproportionate share of partygoers are affiliated with Anthropic or OpenAI, fueling concerns that career pressure on women in AI may push attendance for networking benefits.
Commentary from substack and Bluesky writers frames the group as illustrative of a broader Silicon Valley culture where sexualized networking could influence perceptions of AI safety experts and industry norms.