Anthropic has opened communication with New York City Comptroller Mark Levine, following the city official's public request that the artificial intelligence company postpone its planned initial public offering over safety and governance concerns. The outreach highlights how questions once confined to AI research labs and policy circles are now being pressed by the public officials whose pension money could ultimately be invested in the company.

Speaking on Bloomberg Television on Friday, Levine confirmed that discussions between his office and the AI company had begun, though he declined to provide details about their nature. “It does appear they are under pressure from multiple sides right now,” Levine said.

The development comes as Anthropic prepares for a potential November stock market listing in what could become one of the largest initial public offerings in history, with investors reportedly discussing valuations between $1.5 trillion and2 trillion.

NYC comptroller warns of AI safety risks

In a letter addressed to Anthropic CEO Dario Amodei on Wednesday, Levine urged the company to delay its IPO until it adequately addresses concerns surrounding AI safety, governance, and the rise of increasingly autonomous systems.

Levine also questioned whether a recent voluntary safety agreement between the Trump administration and major AI companies, including Anthropic, provides sufficient safeguards to address the risks posed by increasingly capable systems — a doubt that sits at the heart of a wider debate over whether voluntary commitments are enough to keep pace with the technology's rapid advance.

“If they go public and then there is a catastrophic safety incident, beyond, far beyond what we’ve seen right now, that would be terrible for their stock,” Levine told Bloomberg.

As comptroller, Levine oversees New York City’s public pension funds, which collectively manage more than $325 billion in assets. Some of these funds hold investments that track major stock indexes, potentially increasing their exposure to Anthropic once it completes its public debut. Because index-tracking funds replicate their benchmarks automatically, the company's shares would enter those portfolios once it joins major indexes.

Levine argued that postponing the listing would give Anthropic additional time to establish a governance framework capable of addressing the very risks that researchers within the company itself have identified.

Anthropic prepares for November IPO amid valuation concerns

The discussions come as Anthropic advances preparations for its stock market debut. During a meeting with selected investors on Wednesday, the company disclosed that its third-quarter revenue reached $16.8 billion, representing a 45% increase from the previous quarter.

However, some prospective investors have reportedly discussed a valuation closer to $1.5 trillion, below earlier expectations of approximately $2 trillion, amid concerns about the costs of AI development. The gap captures the tension at the heart of the listing: rapid revenue growth set against the steep costs of developing frontier AI systems.

Meanwhile, Amodei has publicly acknowledged the dangers associated with increasingly capable AI systems. In an essay published last month, he called for greater cooperation between governments and AI developers to address risks including loss of control, economic disruption, cyberattacks, and terrorism — warnings that closely mirror the concerns Levine raised in pressing for a delay.

Despite acknowledging concerns that postponing the IPO could create market uncertainty, Levine maintained that the potential consequences of a catastrophic safety incident occurring after the listing warranted additional caution.

“I worry about going public and then having an incident,” he said.

With a potential November debut approaching and valuation discussions still underway, how the company addresses the safety and governance questions raised in the letter is likely to remain in focus heading into the listing.