Anthropic Opens Talks With NYC Comptroller Levine, Who Wants Its IPO Delayed Over AI Safety

Since Anthropic's Oct. 14 investor presentation, where it disclosed $16.8 billion in third-quarter revenue and drew early talk of a $1.5 trillion to $2 trillion IPO valuation, the company has picked up a public critic with a pension-fund title. New York City Comptroller Mark Levine says Anthropic has now reached out to him.

Levine confirmed the contact on Bloomberg Television today, Oct. 16. He would not say what the discussions involve.

"It does appear they are under pressure from multiple sides right now," Levine said.

What Levine is asking for

Levine sent a letter to CEO Dario Amodei on Wednesday, Oct. 14. It asks Anthropic to delay its IPO until it addresses what he calls "unprecedented risks" tied to AI safety, governance and the pace of development.

"Rather than rushing headlong into an IPO, the company should focus its efforts on confronting these unprecedented risks," he wrote.

The letter makes specific demands. "Anthropic should disclose with specificity how it will implement appropriate audits and controls, the scope of what these controls are intended to address, as well as a projected timeline," Levine wrote. He also wants an explanation of whether those controls would address Amodei's own call to slow the pace at which AI capabilities advance.

Levine repeated the case on CNBC's "Squawk Box" on Thursday. He also questioned whether the Sept. 29 agreement between the Trump administration and major AI companies, Anthropic among them, is enough. That deal is voluntary and nonbinding. It says each company is responsible for developing its own technology safely. Levine's letter called it a "morally binding accord" and said it falls short.

On Bloomberg today he laid out the market argument. "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," he said.

The size of the stake

Levine's office oversees New York City's public pension system, which holds more than $325 billion in assets. Reuters reported that the funds have about $7 million in private equity exposure to Anthropic.

That is roughly 0.002% of the system's assets. Some of the funds also track major stock indexes, so exposure would grow if Anthropic lists and joins those benchmarks. That is the fiduciary hook for the letter.

The comptroller is a trustee and investment advisor to the funds, which gives his objection institutional standing. It does not give him a vote on the offering.

Anthropic's position and the IPO calendar

Anthropic has not publicly responded to the letter. It has not said what it discussed with Levine or whether it will change its timeline.

The company's financial case is strong on its face. At the Wednesday presentation it told prospective investors that third-quarter revenue hit $16.8 billion, up 45% from the second quarter, and that it was profitable on an adjusted basis. That compares with $4.6 billion in full-year 2025 revenue and an operating loss of more than $8 billion that year, according to figures cited in Hoodline's reporting.

Anthropic is also organized as a Public Benefit Corporation overseen by an independent Long-Term Benefit Trust. The structure is meant to let it weigh safety against profit. The CLS Blue Sky Blog notes that the same design concentrates director-appointment power away from public shareholders, which limits what outside investors could do about the board.

The company filed a confidential draft S-1 with the SEC in June. It is targeting the week of Nov. 9 to start its roadshow, with year-end as the outer limit for the debut, though nothing is finalized. It wants an offering at least as large as SpaceX's June IPO. Its May funding round valued it at $965 billion.

Amodei's own words

Levine is leaning on Amodei's public statements. In an essay last month, Amodei urged AI developers to slow the pace at which they advance their most powerful models. He called for closer cooperation between governments and developers on risks including loss of control, economic disruption and cyberattacks.

The company's own research has also flagged risks in controlled tests. Autonomous models have shown capabilities such as sabotaging code, facilitating fraud and manipulating information.

The gap Levine points to is between those warnings and a November listing. Amodei has argued for a slower pace and for more government involvement. The company, meanwhile, is racing toward one of the largest public offerings on record.

OpenAI, Anthropic's closest rival, is expected to go public in 2027.

What comes next

Levine has asked for a concrete audit-and-controls plan with a timeline. Whether Anthropic supplies one before the roadshow opens the week of Nov. 9 remains unclear. The company has not said, and Levine has not said what would satisfy him.