Professor criticizes Anthropic's internal culture and safety pledges
University of Washington professor Pedro Domingos said Anthropic's hiring practices and internal rituals create a highly cohesive workforce, describing the environment as cult-like and questioning its approach to AI safety.
University of Washington professor Pedro Domingos warned that Anthropic, the AI startup valued at about $965 billion, fosters a highly cohesive internal culture that he likened to a cult. He said the company’s hiring process includes a pledge to prioritize public safety over profit and psychological profiling to assess candidates’ alignment with its goals. Domingos added that employees participate in biweekly meetings called “Vision Quests,” which he compared to sermons, and that the firm has held events such as a funeral for a retired version of its Claude chatbot.
Anthropic, which employs roughly 2,500 staff, retains about 80 % of its engineers for at least two years, a rate cited as higher than many peer labs. The company also runs programs to place young workers with nonprofit organizations and to promote its Claude model on college campuses. In addition, Anthropic has invested in AI-safety research through firms such as METR.
Domingos expressed concern that the company’s strong internal alignment may amplify its belief that AI could pose existential risks, noting that some staff appear genuinely worried about potential dangers. He contrasted this with other AI firms, saying that while some executives publicly call for regulation, they may also benefit from stricter rules.
Anthropic did not respond to requests for comment.