Anthropic faces criticism over AI safety ahead of IPO

What happened
Anthropic PBC is moving forward with its initial public offering despite facing criticism from AI safety experts and New York City Comptroller Mark Levine. Levine publicly stated his disappointment and indicated that city pensions are prepared to push for stronger AI safety measures. A group of AI safety and ethics figures released the 'MisAnthropic Statement,' accusing Anthropic of weakening its safety commitments while using safety messaging to attract investors and shape regulation. Some potential supporters of the statement declined to sign due to concerns about damaging their relationship with Anthropic or risking future funding.
Key details
- New York City Comptroller Mark Levine criticized Anthropic's IPO timing and safety approach.
- The 'MisAnthropic Statement' was signed by a range of AI safety and ethics figures, including researchers from Harvard, MIT, and UCLA.
- The statement alleges Anthropic has weakened its AI safety pledges and uses safety branding for investor and regulatory advantage.
- Some individuals and groups declined to sign the statement out of concern for professional or financial repercussions.
- Anthropic is currently using SpaceX's Colossus data center to train its Claude AI model.
Why it matters
Investors, AI researchers, and public pension funds are directly affected by Anthropic's decisions around safety and its IPO. The criticism and public statements increase pressure on Anthropic to clarify or strengthen its safety practices, and may influence how regulators and investors assess the company's risk profile. Pension funds, in particular, may use their influence to push for changes in corporate governance or safety commitments at Anthropic.

