Anthropic Engages NYC Comptroller Amid IPO Safety Concerns

Anthropic, the AI company known for its Claude model family, has begun talks with New York City Comptroller Mark Levine following his request for the firm to delay its planned initial public offering (IPO). Levine expressed concerns over AI safety and governance, urging the company to address these issues before proceeding with its IPO.

In a recent appearance on Bloomberg Television, Levine confirmed that discussions with Anthropic have started but did not elaborate on their specifics. He emphasized the pressure the company is facing from various stakeholders as it prepares for a potential November listing.

Levine's concerns stem from the risks associated with advanced AI systems, which he believes need to be adequately addressed before the IPO. He pointed out that a recent voluntary safety agreement involving major AI companies, including Anthropic, may not provide sufficient safeguards against potential catastrophic incidents.

As Anthropic gears up for what could be one of the largest IPOs in history, with valuations discussed between $1.5 trillion and $2 trillion, Levine's caution reflects growing scrutiny from institutional investors. He oversees public pension funds in New York City, which manage over $325 billion in assets and could be significantly impacted by the IPO.

Despite acknowledging that delaying the IPO might create market uncertainty, Levine argued that the potential consequences of a major safety incident post-listing necessitate a more cautious approach. He stated, “I worry about going public and then having an incident.”

Market participants are closely monitoring Anthropic's disclosures, particularly regarding safety and governance, as these could significantly influence investor sentiment and the IPO's reception.

New Insights from TechCrunch Disrupt 2026

Replit CEO Amjad Masad announced that the company is no longer exclusively working with Anthropic, emphasizing a strategy of multi-model flexibility to avoid reliance on a single AI provider.

Replit has implemented AI agents for initial triage on support tickets since early 2026, allowing employees to focus on strategic decisions.

The company's revenue has skyrocketed from $2.8 million in 2024 to a projected annual run-rate nearing $1 billion by 2026, with a valuation increase from $3 billion to $9 billion following a $250 million funding round.

Masad highlighted the competitive landscape, noting that AI model providers like Anthropic and OpenAI are also developing their own coding tools, which poses a risk to Replit's business model.

Despite the competitive tension, Replit continues to integrate Anthropic's Claude models into its Agent product, which has been central to its growth since 2024.