A full council, under oath
The New York City Council sits today as a Committee of the Whole — all 51 members, a format it has not used since 2022 — to take testimony on the risks of artificial intelligence. Jacob Coxon, who resigned from Anthropic last month, will testify at the request of Council Speaker Julie Menin, Bloomberg reported and Reuters relayed.
Coxon is expected to be joined by two others who left frontier labs: Alex Turner, formerly of Google DeepMind, and Daniel Kokotajlo, a former OpenAI researcher who now runs the AI Futures Project. On leaving Anthropic, Coxon said the people building AI “earnestly believe that it could kill us all by the end of the decade”, and accused his former employer and OpenAI of “gambling with our lives”.
What makes today unusual is not the witnesses but the oath. Congress has held AI hearings for three years without compelling anyone; a city council has now done it.
How the companies got here
Menin invited the chief executives — Dario Amodei, Sam Altman, Sundar Pichai, Elon Musk and Mark Zuckerberg. None are coming. Anthropic, OpenAI, Google and Meta are each sending representatives instead, and they will testify under oath.
The path to that agreement was not voluntary for most of them. Meta agreed without pressure. Google and Anthropic declined by the Council’s Friday deadline and agreed after being warned that subpoenas would follow. OpenAI confirmed on the Sunday, after the same warning. SpaceXAI did not respond at all and was formally served with a subpoena, amNewYork reported. If the company does not comply, the Council can ask the New York State Supreme Court to enforce it.
“These companies have a significant presence in New York City and collectively employ tens of thousands of New Yorkers,” Menin said, adding that they “owe it to their workers and, above all, to the public to come before the Council.”
The bills behind the hearing
The hearing exists to move a package of bills the Council announced on 25 September. The proposals would require AI systems sold in the city to pass third-party validation and to carry a kill switch; pay whistleblowers a portion of the fines their reports produce; and create a private right of action letting New Yorkers sue AI companies when a jailbroken tool harms them.
The broadest measure sets a penalty of $25,000 per instance and requires independent checks covering data quality, bias, privacy and security. A private right of action is the provision industry lobbies hardest against anywhere it appears, because it moves enforcement out of a regulator’s discretion and into the hands of anyone with a lawyer.
Why a city
New York is legislating into a vacuum. There is still no comprehensive federal AI statute, and the current administration’s posture is to preempt state rules it considers burdensome rather than to write its own. A city cannot regulate model training, but it can regulate what is sold within its boundaries, and New York is large enough that compliance there tends to become the default elsewhere.
The jurisdictional question is real and will be litigated if the bills pass. A $25,000-per-instance penalty attached to a product sold nationally invites an argument about where the city’s authority ends.
What to watch is narrower than the bills, though. It is whether sworn testimony produces anything that voluntary testimony has not. Executives have appeared before legislatures repeatedly and said what they chose to say. Today, three people who left these companies will answer the same questions, under the same oath, on the same record.
