PromptCrates
Policy & RegulationPolicy & Regulation 6 min read

NYC Council Grills OpenAI, Anthropic, Google and Meta Under Oath

The New York City Council questioned OpenAI, Anthropic, Google and Meta under oath on 5 October 2026 as it weighs local AI safety bills, while SpaceXAI ignored a subpoena.

PC

PromptCrates Editorial

Staff Writer

0 0
NYC Council Grills OpenAI, Anthropic, Google and Meta Under Oath

The New York City Council on Monday, 5 October 2026, questioned representatives of OpenAI, Anthropic, Google and Meta under oath about the risks of advanced AI, convening all 51 members as a Committee of the Whole while it weighs a package of local AI safety bills. Elon Musk's SpaceXAI, the fifth company called to testify, did not appear despite a council subpoena.

Speaker Julie Menin led the questioning. According to CNBC, the companies sent Logan Graham, head of Anthropic's Frontier Red Team; Morgan Dwyer, OpenAI's head of policy development and operations; Alice Friend, Google's director of AI and emerging tech policy; and Shane Cahill, Meta's AI policy director for legislation. CNBC reported that the four appeared by video link, and amNewYork said roughly 40 council members took part over the day, citing the Council.

What the NYC Council asked the AI companies

Menin opened by asking each company to put a number on the chance of a catastrophic outcome. amNewYork reported that Dwyer answered "I don't know" and argued that whether the risk was 1%, 10% or 20%, "none of these levels is remotely acceptable." Menin called that answer "flippant at best." Graham described Anthropic's risk assessment work without giving a percentage, Cahill said he would follow up, and Friend said there was not yet a rigorous scientific method for assigning a probability to such an event, according to the same report.

CNBC quoted Menin concluding that none of the four companies could quantify the risk of something cataclysmic, which she called "troubling at best." She also asked whether failing an internal safety test or an independent validation would stop a model's release. amNewYork reported that the companies described their review processes without giving the blanket commitment she wanted, and that Dwyer said OpenAI would not release models it did not believe were safe.

On legal liability, Friend repeated Google's position that "if it's illegal without AI, it's still illegal with AI," CNBC reported. Dwyer said OpenAI was responsible for developing and evaluating its systems safely, while Graham and Cahill said the question fell outside their areas, according to amNewYork.

OpenAI also told the council it is reviewing possible past misalignment incidents involving its AI agents going back to November 2025. Asked whether any had affected a city system or a resident's data, Dwyer said she was not aware of one and committed to informing third parties if New York City was affected, amNewYork reported. The question follows months of disclosures, including the official report OpenAI released on the Hugging Face incident.

Former lab researchers warn of lost control

The company panel followed testimony from three former insiders. Jacob Coxon, a former researcher at OpenAI and Anthropic, told the council that on the current path it is "more likely than not that humanity loses control" of these systems, and amNewYork reported that he framed this as his own assessment. Former OpenAI researcher Daniel Kokotajlo called for more transparency and a slowdown in leading model development, according to CNBC.

Alex Turner, a former Google DeepMind researcher, warned that "we are racing to build and grow our own adversary here at home, which is misaligned AI," CNBC reported. amNewYork said Turner put his personal estimate of an eventual AI takeover at roughly one in three.

City & State New York counted 14 panels across the marathon session. Menin told the outlet around the fourth hour that it was "disappointing that they can't seem to answer some of the most basic, rudimentary questions."

The local AI bills behind the hearing

The Council tied the hearing to legislation. Its press release of 28 September said the proposals include a whistleblower incentive program, a private right of action for New Yorkers harmed by AI agents and independent third-party validation requirements. amNewYork added that the package includes human "kill switches" and 24-hour reporting of certain safety incidents involving city agencies or contractors.

PPC Land, which reviewed the committee's briefing paper, reported that the hearing covered two introduced measures and eight preconsidered drafts. By its account, Menin's validation draft would bar selling or deploying an AI model in the city unless a third-party validator had checked it and it had a shut-down capability, with a civil penalty of $25,000 per instance and an effective date 180 days after enactment. A companion draft would let residents file complaints and receive 25% of any proceeds the city recovers, PPC Land reported.

The hearing's Legistar record lists it as an oversight item titled "Examining the Risks Posed by Artificial Intelligence," on the 5 October agenda. PPC Land said none of the ten measures has passed and the documents it reviewed give no vote date.

SpaceXAI, the RAISE Act and Washington

The 28 September press release said Meta agreed to appear voluntarily, that OpenAI, Google and Anthropic confirmed only after the Council warned it would issue subpoenas, and that Menin subpoenaed SpaceXAI that day. At the hearing, Menin said SpaceXAI's absence was "in direct violation of the subpoena" and that the Council would pursue the matter in court, CNBC reported.

State lawmakers used the session to revisit New York's RAISE Act. State Sen. Andrew Gounardes testified that the companies spent millions of dollars to defeat the bill's original version, and Assembly Member Alex Bores later wrote on X that he believed OpenAI "just committed perjury," according to amNewYork. Dwyer told the council that OpenAI supports the RAISE Act that passed.

Menin opened the day by criticizing what CNBC described as Washington's light-touch approach, after President Donald Trump signed a voluntary accord with tech leaders the previous week, covered here as the White House super intelligence accord. PromptCrates also covered the Australian Senate's call for Altman and Amodei to appear after rogue-agent reports.

Mayor Zohran Mamdani said AI needs to be regulated "at every single level" of government but that the most urgent response must come from Washington, and that he wanted to review the bills before backing them, amNewYork reported. Menin said the companies will receive written questions on the points they did not answer.

policy-regulationNew York City CouncilAI safetyOpenAIAnthropiclocal AI regulation

Related articles