OpenAI, Google, Meta, Anthropic face legal questions at NYC hearing

Four of the most powerful AI companies on the planet sat down in front of the New York City Council and took questions under oath. The questions were about who pays if things go catastrophically wrong.

On October 5, 2026, the council convened a Committee of the Whole hearing, a format in which the full legislative body sits together as a single committee. Executives from OpenAI, Google, Meta, and Anthropic gave sworn testimony on AI safety and the potential for catastrophic outcomes. The hearing ran alongside discussion of 10 local AI-related bills aimed at tightening oversight of the labs building these systems.

Menin wanted numbers. She got hedging.

Council Speaker Julie Menin led the questioning, and her line of attack was simple. Can these companies actually quantify the risks they are talking about?

One risk she focused on was loss of control over AI models. That is the scenario where a system stops doing what its builders intended, and nobody can easily pull it back.

Menin also asked a pair of follow-ups that cut to the financial heart of the matter. Would these companies accept legal responsibility for a catastrophic incident? Would they provide insurance to cover one?

The executives’ responses were largely non-committal. Menin sharply criticized the representatives, particularly over what she saw as the dismissal of risk quantification as inconsequential.

OpenAI’s Morgan Dwyer did offer one firm position. Dwyer stated that no level of catastrophic risk is acceptable, while calling the responses to risk quantification “flippant.”

Whistleblowers bring a darker forecast

The executives were not the only people testifying. Whistleblowers from these companies also spoke, raising serious concerns about industry practices and indicating that oversight may be inadequate as AI systems become more deeply woven into everyday life.

The starkest claim came from whistleblower Jacob Coxon. He put the risk of AI leading to human extinction at greater than 50%.

Why a city council is doing this

New York City is moving with 10 AI-related bills that were introduced or discussed in connection with the hearing. Collectively, they target accountability for AI labs and the safety protocols those labs follow. The hearing was positioned as a potential template for future regulations in the absence of overarching federal guidelines, with emphasis on accountability mechanisms including independent validation and liability pathways.

The Committee of the Whole format signals that the council treated this as a body-wide priority rather than a niche item for a single subcommittee. Sworn testimony carries more weight than a press release or a blog post, and executives who chose vague answers in that setting made a deliberate choice about what they were willing to put on the record.

What this means for the AI industry

Liability and insurance are the language of accountability that every other high-risk industry already speaks. Menin’s questions effectively asked whether AI labs should be treated the same way. The non-committal responses suggest the companies are not eager to volunteer for that treatment.

The next thing to watch is how the 10 bills progress, particularly measures on accountability, safety protocols, third-party validation, and whistleblower incentives. Menin asked for risk numbers, a liability commitment, and an insurance commitment. A follow-up from any of the four labs on those points would say a lot about how seriously they are taking local oversight.

For now, the record shows a city government asking clear questions about who pays when AI fails catastrophically. The companies building that AI have not yet given a clear reply.