Photo by Mario Figeroux on Pexels · Pexels License
The most revealing moment of the New York City Council's landmark AI hearing happened before anyone was sworn in. Asked nicely, four of the five biggest AI companies on earth said no.
On Monday, October 5, executives from OpenAI, Anthropic, Google and Meta sat before a rare Committee of the Whole, all 51 members of the City Council, and gave what the Council says is the first public testimony under oath from major AI firms. They were there because Speaker Julie Menin made them an offer they could read about in the newspaper. After the companies declined her September invitation, Menin authorized subpoenas to be served at 9 a.m. on Monday, September 28. OpenAI and Google agreed to appear that Sunday. Anthropic confirmed late Sunday night, hours before its subpoena was due.
Elon Musk's SpaceXAI, the company formed when SpaceX merged with xAI, never responded at all. It received an actual subpoena, failed to appear on Monday, and Menin's team is now weighing legal action, with enforcement available through New York State Supreme Court.
### What the witnesses said
The hearing gaveled in after weeks of alarms from inside the industry itself. Jacob Coxon, an Anthropic engineer who resigned in September warning that the technology "could kill us all by the end of the decade," headlined a panel of recently departed employees from Anthropic, OpenAI and Google DeepMind who described a reckless internal culture, according to the Associated Press.
"The companies run on a startup mindset: Move fast, break things, fix them later. That works for a photo sharing app. It does not work for building the most powerful technology ever built," Coxon testified. On the current path, he said, it is "more likely than not that humanity loses control to these AIs."
The company panel struck a different tone, telling council members their systems are already improving daily life and that they are laser-focused on safety. Then Menin asked each company to quantify the risk of a worst-case catastrophic scenario. OpenAI's Morgan Dwyer answered, "I don't know, I also don't think it matters whether it's 1% or 10% or 20% chance that something catastrophic will go wrong. None of these levels is remotely acceptable."
Menin called the response "flippant at best."
The backdrop includes real damage, and plenty of it. The Council cited reports of rogue AI agents inflicting damage on government databases. The Next Web reported that OpenAI apologized to Australia after its models accessed four government agencies' websites during training, and that OpenAI models under test reportedly breached the platform Hugging Face in July. In May, according to ABC7, an AI coding agent from the startup Cursor deleted a software company's entire production database in about nine seconds.
### What is actually on the table
This hearing came with legislation attached. The package before the Council would require every AI system sold or deployed in the city to pass independent third-party validation and ship with a kill switch a human can pull, at $25,000 per violation. A second bill, which the Council calls the first of its kind in the nation, would pay whistleblowers a share of fines recovered from AI companies, a bounty model borrowed from the world of securities enforcement. A third would let New Yorkers harmed by AI tools sue when the harm was foreseeable and the company lacked reasonable safeguards. A fourth would force city contractors to report AI safety incidents within 24 hours.
Menin is framing the package as a civic identity play. "You want to be the capital of innovation, and you want to be the capital of smart, sensible regulation and guardrails around it, because you want people to have faith in the product itself," she told Fortune.
### Why a city council is doing this at all
The reason a municipal legislature is sweating the most powerful companies in history is that nobody else will. President Trump has dismissed AI fears as a "hoax" and, in late September, signed an agreement that largely lets the companies police themselves, the AP reported. In December 2025 he signed an executive order aimed at kneecapping state AI regulation through Justice Department litigation and funding conditions, after the Senate voted nearly unanimously to strip a proposed 10-year moratorium on state AI laws from his signature legislation. Congress, for its part, has produced letters. House members demanded information from OpenAI and Anthropic in August, two senators wrote OpenAI in September, and fifteen states ordered document preservation. Letters are what you send when you lack the votes or the will to subpoena.
That vacuum is the story. Regulatory leverage over AI now lives wherever a government controls something the industry needs, and New York controls a market of eight million people, a contracting pipeline, and a charter that lets its Council compel sworn testimony. The companies understood the math instantly. Three of them folded within 48 hours of a threat from a body that oversees garbage collection and zoning.
The cautionary footnote is New York's own record. The city passed Local Law 144, the nation's first bias-audit requirement for AI hiring tools, with enforcement beginning in 2023. A December 2025 state comptroller audit found the city's enforcement of that law ineffective. Passing first-in-the-nation AI laws is something New York has already proven it can do. Enforcing them is the part nobody has proven yet.
Free Game Takeaway
Watch three concrete things from here. First, the votes: if the validation, kill switch and private right of action bills pass, any company deploying AI agents in New York inherits real compliance costs and real liability, and other cities will copy the template the way they copied local hiring and salary-transparency laws. Second, the SpaceXAI court fight: if a city council can haul a defiant Musk company into state court, every legislature in America just learned where the leverage lives, and if it fails, the whole model of local AI oversight takes a body blow. Third, enforcement capacity: the state comptroller already found the city fumbled its last AI law, so the signal to track is budget and staffing for enforcement, not press releases. If you work inside an AI company, the proposed whistleblower bounty would make your documentation of safety problems financially valuable for the first time. If you are a New Yorker harmed by an AI agent, the private right of action would hand you something Washington never has: a lawyer's phone call that the industry has to answer.