Anthropic’s IPO puts its safety mission under scrutiny

Anthropic has filed its IPO prospectus, putting its commercial ambitions alongside its stated commitment to developing powerful AI responsibly. The company’s public messaging portrays its technology as potentially dangerous and positions Anthropic as especially equipped to manage those risks. That stance raises questions about the competing priorities of a safety-focused organization and a profit-driven business.

Cofounder and policy chief Jack Clark told an IMF meeting in Bangkok that the technological singularity had begun. Fellow cofounder Chris Olah previously told Vatican officials that Anthropic’s systems showed evidence of introspection and internal states resembling human emotions. Such claims remain contested, with many researchers skeptical of suggestions that software has humanlike inner experience.

Nathan Calvin, chief strategy officer at regulation advocacy organization Encode AI, challenged Anthropic’s trustworthiness after it privately circulated draft legislation to US lawmakers that was subsequently leaked. CEO Dario Amodei defended a “race to the top” strategy, arguing that responsible development can raise industry standards and works better when the company remains at the technological frontier. The dispute highlights unresolved questions about self-regulation, independent safety testers’ access to models, and whether Anthropic’s policy positions serve public safety or its own interests.