New York City Comptroller Mark Levine has asked Anthropic to delay its planned U.S. initial public offering until it addresses concerns about safety, governance and the pace of AI development, Reuters reported. The letter to Chief Executive Dario Amodei ties a frontier AI developer's stock market debut directly to how it manages the risks of its technology, bringing AI safety into the language of investor stewardship.

What Happened

Levine, who oversees the city's public pension funds, wrote to Amodei urging the company to hold off on its offering, according to Reuters, which saw the letter Thursday. CNBC first reported the letter's contents earlier that day.

"Instead of proceeding headlong with its IPO, the company should focus its energies on addressing these unprecedented risks," Levine wrote, according to Reuters.

RelatedThree Fired OpenAI Safety Researchers Give Their Side in Open Letter, Deny Leaking

The comptroller's office told Reuters the city's pension funds have about $7 million of exposure to Anthropic through private equity holdings. The funds manage more than $325 billion in total, which puts the Anthropic stake at roughly 0.002% of assets.

Anthropic did not immediately respond to Reuters' request for comment.

Why It Matters

The financial stake behind the letter is tiny. Its significance lies in who is making the argument. Public pension funds are among the large, long-term investors that buy shares in big listings and vote on governance once a company is public. A comptroller framing AI safety as a reason to wait signals that some of those investors may treat model risk the way they treat climate or labor risk: as a governance issue that belongs in disclosures and board oversight, not only in a company's research papers.

That matters beyond Anthropic. Boards of companies that build or rely on frontier models may face similar questions from shareholders about how AI risks are identified, overseen and reported. Reuters noted that AI has advanced rapidly, outpacing regulation and drawing scrutiny over safety, cybersecurity and misuse, and that Anthropic's own research has found increasingly autonomous models showing potentially harmful capabilities in controlled tests, including sabotaging code, facilitating fraud and manipulating information.

Context

Anthropic is reportedly targeting a valuation of about $2 trillion, Reuters said, and at a pre-IPO meeting this week told prospective investors that third-quarter revenue rose 45% from the prior quarter, according to sources cited by Reuters. Rival OpenAI is expected to list in 2027.

Bloomberg has reported that Anthropic aims to start marketing the IPO as soon as the week of Nov. 9. Its prospectus devotes extensive space to risk factors, including behaviors observed in its own models, and its founders have sought voting control ahead of a listing, a structure that would limit how much outside shareholders can influence the company. Anthropic has also received an FTC subpoena over AI safety incidents, according to Axios.

What to Watch

Whether Anthropic responds publicly to Levine, and whether other public pension systems or proxy advisers raise the same concerns. Investors will also watch how the company addresses safety oversight and founder control during its roadshow, and whether the timing of the offering shifts.