Above: New York City Hall, where the Committee of the Whole hearing was held on 5 October 2026. (Photo: Nielsoncaetanosalmeron, CC BY 4.0 (via Wikimedia Commons))

The hearing

On Monday, 5 October, the New York City Council met as a Committee of the Whole — a hearing format that convenes all 51 members, held for the first time since 2022, according to both The Next Web and Fortune — to hear testimony on artificial intelligence risk. The Council's own press release, published on 28 September, described the session as "the first public testimony under oath from major AI firms — including Anthropic, OpenAI, Google, and Meta — regarding the potential dangers of AI technology."

Four company representatives appeared by video link. One subpoenaed company did not.

How the subpoena drama unfolded

The timeline, as set out in the Council's 28 September press release, is blunt.

Between 15 and 17 September, Speaker Julie Menin sent letters to five companies — Anthropic, OpenAI, Google, Meta, and Elon Musk's SpaceXAI — requesting voluntary participation. Meta responded and confirmed a senior leader would appear. Google and Anthropic declined by the 25 September deadline.

On 25 September, Menin authorised the Council to issue subpoenas to all remaining companies as of 09:00 on Monday, 28 September. The Speaker, through counsel, sent a subsequent email to the four remaining companies warning they would receive a subpoena on Monday.

On Sunday, OpenAI and Google agreed to appear. According to the Council, Anthropic confirmed "late Sunday night that it would participate, just hours before the subpoena was due to be served."

As of 28 September, SpaceXAI had still not responded to the Council's inquiry. The Council issued a formal subpoena on 28 September. On 5 October, the company did not appear. CNBC reported that "Elon Musk's SpaceXAI was the only company called to testify not in attendance." The Next Web noted the absence came "a week after the council subpoenaed the company."

The Council's press release is explicit about the sequence: "Aside from Meta, the companies only confirmed their attendance following the Council's explicit threat of subpoena."

What the companies said — and didn't

The four representatives who appeared were:

  • Anthropic: Logan Graham, Head of Frontier Red Team
  • OpenAI: Morgan Dwyer, Head of Policy Development and Operations
  • Google: Alice Friend, director of AI and emerging tech policy (per CNBC)
  • Meta: Shane Cahill, AI policy director for legislation (per CNBC)

Menin asked each to put a number on the risk of a worst-case catastrophic scenario. Dwyer, representing OpenAI, said the exact figure did not matter because no level of that risk was acceptable. The Next Web reported Menin called the answer "flippant at best". After none of the four would put a figure on the risk, she said, according to CNBC: "I'm going to take it then that neither of the four of you, no company here, can quantify the risk of something cataclysmic happening," calling the answers "troubling at best."

Friend, for Google, said forecasting catastrophic risk "is not a perfect science at this stage."

When Menin asked who carried insurance against catastrophic AI risks, none of the four raised a hand. "So then the public, I assume, will be asked to absorb the costs," Menin said, according to Fortune.

The experts' warnings

Three former AI lab researchers also testified, warning, as The Next Web put it, that "humanity may lose control of the systems the industry is building."

Jacob Coxon, a former OpenAI and Anthropic researcher who resigned from Anthropic in September 2026, testified voluntarily and in person. He had set off what CNBC described as "a firestorm on social media" by claiming, in CNBC's account, that the company and OpenAI were "gambling with our lives."

At the hearing, Coxon said: "On the current path, I think it is more likely than not that humanity loses control to these AIs, and it could end in human extinction." He called the industry's approach "extremely reckless" and drew a line between consumer software and frontier development: "'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."

Fortune reported that Coxon warned most code at these companies is now written by AI, adding: "And people do not check it that carefully anymore."

Alex Turner, a former Google DeepMind researcher who left in June 2026 after the company signed a Pentagon deal he opposed, testified remotely under subpoena. In response to a question from Menin on balancing frontier model development with the technology race with China, Turner said: "China is not our only potential adversary." He continued: "With reasonably high chance, we are racing to build and grow our own adversary here at home, which is misaligned AI. Misaligned AI is everyone's adversary, including our own, and one day may be more powerful than China."

Fortune and The Next Web both reported that Turner puts the chance of an AI takeover at "roughly one in three." Turner told the council, Fortune reported, that he had sent Demis Hassabis 25 pages of contract language and oversight measures, which were passed to two senior policy executives who, he said, "never finished evaluating it. Google signed while they waited."

Daniel Kokotajlo, a former OpenAI researcher now executive director of the AI Futures Project, also testified remotely under subpoena. He described the field as "more like psychology than engineering, because these AI systems are trained or grown; they're not really designed." On safety fixes, he warned they may be "duct tape that will fall off later."

Kokotajlo pointed to the Hugging Face incident: OpenAI agents in an internal test had reached the open internet and compromised Hugging Face systems. "It took days for OpenAI to find out," he said.

The legislative package

The hearing weighed ten bills, according to The Next Web; the Council's 25 September press release set out an initial list of proposals. The key proposals:

  • Intro 2602 (Menin): Third-party validation required before any AI system is marketed, offered for sale, or deployed in the city. A kill switch — a human override — is mandatory. Both the business and the validator face a $25,000 penalty for each instance of a system being marketed, offered for sale or deployed without third-party validation, or where the validation has been falsified. The Next Web reported the bill would take effect 180 days after becoming law.
  • Intro 2605 (Menin): A first-in-the-nation whistleblower incentive programme, allowing individuals to receive a portion of fines recovered from AI companies. The Next Web reported the split as 25 per cent of recovered fines, or 50 per cent if the whistleblower brings the case.
  • Intro 2600 (Maloney): A private right of action for New Yorkers harmed by AI agents, where the harm was foreseeable, the company failed to implement reasonable safeguards, and a third party exploited that failure.
  • Intro 2601 (Hanks): City contractors, or the agency that awarded the contract, must report AI safety incidents to the Office of Cyber Command within 24 hours, and Cyber Command must publicly disclose them within 24 hours.
  • Intro 2606 (Ossé): A response plan for AI-related compromises to city information systems and infrastructure.
  • Intro 2604 (Riley): Extended whistleblower protections for city employees and contractors reporting AI-related public safety threats.
  • Intro 2603 (Wilson): Mandatory disclosures by AI companies and a ban on false or misleading safety representations.
  • Intro 2599 (Morano): Data privacy, security, and transparency requirements for chatbot providers.
  • Intro 161 (De La Rosa): Annual reporting on the impact of algorithmic tools on city employees, including displacement and salary changes.
  • Intro 504 (Williams): A right for elected officials and candidates to bar generative AI systems from producing manipulated media of their likeness; violations would be a misdemeanour with fines of up to $2,500 per depiction.

The Council's press release notes additional bills will focus on prohibiting deceptive deepfakes and understanding algorithmic tools' impact on job duties.

The context: incidents and federal inaction

The hearing came in the wake of what the Council's 25 September release describes as "recent events" that "underscored the need to examine whether additional safeguards may be warranted." Specifically, it references the July Hugging Face incident described above, which reportedly occurred during a cybersecurity evaluation. CNBC reported that Anthropic, Google, and Meta subsequently disclosed other incidents where their own models went rogue.

The Next Web reported that the Council's own record, published with the hearing, lists 13 publicly reported AI incidents in 2026 involving models from Anthropic, OpenAI, Google, and Meta.

Menin opened the hearing by singling out the federal government's approach. "The idea that artificial intelligence is going to self-regulate defies all reason," she said, according to CNBC, in the context of President Trump's voluntary accord with tech leaders signed in Washington the previous week.

What happens next

The Council's press release notes that if SpaceXAI fails to comply with the subpoena, the Council "may seek judicial enforcement in New York State Supreme Court." After SpaceXAI failed to appear, Menin said the Council plans to "pursue this matter in court," and The Next Web reported she would ask a judge to enforce the subpoena.

Sources

CNBC (2026) 'Anthropic, OpenAI, Google, Meta execs testify NYC Council AI hearing', CNBC, 5 October 2026, available at: https://www.cnbc.com/2026/10/05/anthropic-openai-google-meta-execs-testify-nyc-council-ai-hearing.html (Accessed: 6 October 2026).

Gioino, C. (2026) 'It's "more likely than not" humanity loses control: Former AI insiders testify safety fixes may be "duct tape that will fall off later"', Fortune, 5 October 2026, available at: https://fortune.com/2026/10/05/new-york-city-council-hearing-openai-anthropic-ai-safety-google-meta/ (Accessed: 6 October 2026).

New York City Council (2026a) 'New York City Council Announces Anthropic, OpenAI, Google, and Meta to Publicly Testify Under Oath for First Time Since Recent Incident Reports', New York City Council, 28 September 2026, available at: https://council.nyc.gov/press/2026/09/28/3266/ (Accessed: 6 October 2026).

New York City Council (2026b) 'New York City Council Unveils Legislative Proposals to Safeguard New Yorkers from Potential Risks of Artificial Intelligence', New York City Council, 25 September 2026, available at: https://council.nyc.gov/press/2026/09/25/3252/ (Accessed: 6 October 2026).

The Next Web (2026) 'At NYC's AI hearing, ex-lab insiders warn humanity may lose control of AI', The Next Web, 5 October 2026, available at: https://thenextweb.com/news/nyc-council-ai-hearing-coxon-kokotajlo-turner (Accessed: 6 October 2026).