New York City Comptroller Mark Levine has urged Anthropic to pause its planned initial public offering until the company addresses concerns about artificial intelligence safety, governance and the pace of AI development.

In a letter to Anthropic CEO Dario Amodei published on October 14, Levine argued that the company cannot warn the public about AI’s dangers while simultaneously racing to the public markets. The request comes as Anthropic pursues a valuation near $2 trillion.

“I write to urge Anthropic to pause its planned initial public offering (IPO) until the company fully and transparently addresses the artificial intelligence (AI) safety, governance, and development pace concerns it has raised,” Levine wrote.

Levine cites Amodei’s call to slow AI development

Levine’s argument relies in part on Amodei’s essay “We Must Pace the Frontier”, published on September 12. In the essay, Amodei wrote, “We must slow the pace at which we improve the capabilities of AI models.”

Amodei said that recursive self-improvement and a cross-industry security incident had persuaded him that the AI industry needed greater caution, rather than simply increased spending on safety.

The essay proposed a three-step plan. The first step would give third-party evaluators employee-like access to verify companies’ safety practices. Anthropic has said it has already committed to that measure independently. The second step called for coordination among AI companies in democratic countries, while the third advocated global coordination, including with authoritarian governments.

According to Levine, Anthropic’s IPO could interfere with the work Amodei described as urgent. The essay received support from parts of the AI industry, with OpenAI CEO Sam Altman and xAI CEO Elon Musk agreeing to its submissions.

Levine is also directly connected to New York City’s investment interests. As comptroller, he serves as investment adviser, custodian and trustee for the city’s pension funds, which hold more than $325 billion in assets. The comptroller’s office said the public pension funds have approximately $7 million of exposure to Anthropic through private equity.

Although that stake is relatively small, Levine said it gives his office a basis for pressing Anthropic before the company becomes publicly traded.

Concerns over Anthropic’s governance structure

Levine also raised concerns about Anthropic’s corporate structure. The company is a Delaware public benefit corporation with a Long-Term Benefit Trust. A public benefit corporation is a for-profit structure that requires directors to balance shareholder returns against a stated public benefit purpose, while a trust of this kind is designed to give an independent governance body a role in safeguarding the company’s long-term mission. According to Levine, reports indicate that the structure could give founders and trustees concentrated control.

He argued that a gap between that control and accountability to shareholders would be difficult to accept given the risks Anthropic has already disclosed.

Levine also criticized a recent agreement between the AI industry and the US government. Anthropic and five other companies signed what the letter described as a morally binding accord with the White House. Levine said the agreement “is insufficient to meet the moment.”

Anthropic’s own disclosures have reinforced his concerns. In its confidential draft S-1 filing, the company warned that advanced AI could pose “catastrophic or existential risks to humanity.” The risk-factors section reportedly runs about 80 pages, nearly twice the length of the section describing the business.

Levine noted that Amodei’s essay was published after Jacob Coxon, a former Anthropic researcher, resigned on September 9. Coxon cited existential concerns as the reason for his departure and said insiders believed AI could wipe out humanity by the end of the decade if development continued at its current pace.

Evan Hubinger, Anthropic’s alignment science lead, has put the odds of AI killing all humans within the decade above 10%.

Anthropic reports rapid revenue growth

Anthropic’s safety and governance debate comes as the company prepares for a potential listing that could value it at more than $2 trillion.

At a private-IPO meeting in San Francisco on Wednesday, October 14, Anthropic told investors that third-quarter revenue reached approximately $16.8 billion, a 45% increase from the previous quarter. The company also said it had generated an adjusted profit.

Anthropic reported revenue of $4.7 billion in the first quarter of the year. That figure increased substantially in the second quarter, when revenue reached approximately $11.6 billion. Despite the growth, the company still recorded an operating loss of more than $8 billion over the period.

Anthropic also told investors that it was cash-flow positive as it exited the third quarter, an unusual claim in an industry whose leading companies have continued to spend heavily on infrastructure and computing.

The company currently has more than $518 billion in planned infrastructure and computing commitments over the next decade. Those commitments form part of the case for a listing that could exceed $2 trillion and potentially bring Anthropic to the public markets before rival OpenAI, which is targeting an early 2027 debut.

Anthropic confidentially filed its draft S-1 registration statement with the Securities and Exchange Commission on June 1. A confidential draft S-1 allows SEC staff to review a company’s registration statement before it is made public, and a public filing is required before shares can be offered to outside investors.

Anthropic has not disclosed an exact IPO date. Speculators have suggested that the offering could take place after the November US midterm elections, although the company is still expected to list before OpenAI. Anthropic’s response to Levine’s letter, and the timing of its move to a public S-1 filing, are now among the key points to watch as the listing process advances.