NYC Is About to Put OpenAI, Anthropic, Google and Meta Under Oath

At 11 a.m. on Monday, Oct. 5, all 51 members of the New York City Council will sit as a single committee and put questions to the companies building frontier AI. OpenAI, Anthropic, Google and Meta have agreed to testify under oath. The fifth company on the invitation list did not answer, so Speaker Julie Menin subpoenaed it on Sept. 28.

That company is SpaceXAI, Elon Musk’s AI venture. If it doesn’t show, the council plans to seek enforcement in New York State Supreme Court and, failing that, could pursue civil contempt sanctions, City & State reports. SpaceXAI did not immediately respond to the outlet’s request for comment.

The hearing is formally titled “Oversight – Examining the Risks Posed by Artificial Intelligence,” according to the BetaNYC civic-tech newsletter, and it is a Committee of the Whole, a format AI Weekly says the council hasn’t used since 2022. It is also the first public test of a 10-bill package Menin unveiled on Sept. 25.

What’s on the table

The centrepiece bill, as 6sqft summarizes it, would require third-party validation of AI systems before they can be marketed, sold or used in the city, with validators checking data quality, bias, privacy and security. Every system would need a “kill switch,” a human override. Violations would cost US$25,000 (CA$34,750) each.

Around it sit the more novel pieces:

• A bounty for whistleblowers. People who report violating AI companies would get a share of recovered fines, which Fortune notes could be the first such law in the country.
• A right to sue over jailbreaks. One bill would let people sue AI companies for harm caused by third parties who get around safety controls, “provided the harm was foreseeable and the company lacked reasonable safeguards,” per Fortune.
• A 24-hour clock. City contractors would have to report AI safety incidents to the city’s Cyber Command within a day, BetaNYC says.
• Protection for city staff who report AI threats to public safety, plus rules on chatbot privacy and false safety claims.

Menin’s pitch is blunt. “This is not an industry that should self-regulate,” she told Fortune. On jurisdiction, her argument is simple: “These companies have offices in New York. The product is being sold in New York.”

What an oath buys

Testifying under oath means a false statement carries legal risk that a press release does not. That’s the real lever here. A city council can’t license a model or halt a training run. It can make a company’s representative answer specific questions on the record, with consequences for lying.

Who those representatives are matters. The council invited the CEOs: Dario Amodei, Sam Altman, Sundar Pichai, Musk and Mark Zuckerberg. Sources told Fortune that none of the five is likely to testify. Expect policy staff, who will be fluent in what their company has published and careful about everything else.

Menin framed the oath as a public obligation. “Given the high stakes, these firms owe it to their workers and, above all, to the public to come before the Council, answer our questions, and provide input on our proposed legislation under oath,” she said, per Route Fifty.

The preemption problem

Even if the bills pass, they walk into a federal fight. Fortune lays it out: President Trump created a Justice Department task force to sue states over their AI laws. “Its first target, Colorado, gutted its own law five weeks after the suit was filed.” A bipartisan bill from Sens. Ted Cruz, Amy Klobuchar and John Thune would likely override most state AI laws, New York’s RAISE Act included. A city ordinance sits below all of that. The jailbreak bill may also test whether Section 230 covers AI output, which courts haven’t settled.

Washington is arguing with itself, too. On Oct. 1 Axios reported that Sens. Josh Hawley and Chris Murphy are pushing an AI Agent Accountability Act to make companies criminally and civilly liable for hacking incidents caused by their AI agents. The administration prefers self-regulation; Axios quotes Director of National Intelligence Jay Clayton: “We have consumer protection laws. We have product liability laws. We have the Department of Justice.”

What to watch Monday

• Whether SpaceXAI appears. An empty chair turns a policy hearing into a court filing.
• Whether anyone commits to a kill switch or explains, on the record, why a city can’t require one.
• Incident disclosure. The 24-hour reporting bill invites a direct question: how many safety incidents has each company logged, and who was told?
• Preemption talk. Listen for which companies argue the city lacks authority. That argument previews the lawsuit.

The public can weigh in. In-person testimony needs no pre-registration, virtual testimony does, and written testimony is accepted up to 72 hours after the hearing ends, according to BetaNYC. BetaNYC’s own director, Noel Hidalgo, is testifying and has been asking New Yorkers a fair question: “What should the Speaker and the Council ask the Mayor’s team and AI companies?”