Former Anthropic researcher Jacob Coxon, fellow whistleblowers, and policy officials from four major AI labs testified under oath at a New York City Council hearing on Monday2,6. ANALYSIS The session establishes a city-level regulatory playbook that, if codified into law, could propagate to other large U.S. municipalities faster than any federal bill now in play.
Why it matters
New York's Committee of the Whole hearing convened all 51 Council members to question representatives from Anthropic, OpenAI, Google, and Meta on the risks posed by AI9. The format, pairing industry insiders-turned-critics with the companies they left, is without precedent at the municipal level. ◆ It matters because the federal vacuum on frontier AI safety has persisted even as New York State passed one of the country's first frontier AI safety laws last year10,11, and the city is now moving to fill the gap with its own ten-bill package.
The big picture
Jacob Coxon, who quit Anthropic last month, delivered the hearing's starkest language. "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 told the Council. "From my experience, the companies are being extremely reckless given the stakes". He was joined by former OpenAI researcher Daniel Kokotajlo, executive director of the AI Futures Project, and former Google DeepMind researcher Alex Turner8. Both Coxon and Kokotajlo warned that the "move fast and break things" culture at the labs is dangerous when applied to frontier AI.
Coxon framed the stakes in terms he had already tested on a national audience: his resignation post accusing Anthropic and OpenAI of "gambling with our lives" surpassed 170 million views on X, and his subsequent television appearances describing superintelligence development as akin to "summoning an alien species" prompted more than 20 members of Congress to call for federal AI guardrails4.
Three of the four companies, OpenAI, Anthropic, and Google, agreed to appear only after Council Speaker Julie Menin's office warned they could be subpoenaed; Meta had committed earlier. "Whistleblowers are why the world learned of the potential dangers of AI, and they deserve to be protected, rewarded, and heard," Menin said.
Between the lines
The ten bills before the Council sketch a regulatory architecture that goes well beyond disclosure. One proposal would require AI systems to include a "kill switch" allowing a human operator to shut them down. Another would create a first-in-the-nation whistleblower program for municipal employees, contractors, and subcontractors who report suspected wrongdoing related to AI development. A separate measure would give whistleblowers a portion of fines or penalties recovered from AI companies. Violations could carry fines of up to $25,000. Yet another proposal would require the city to track and report how AI is affecting its workforce, including jobs eliminated or displaced.
ANALYSIS The combination of a kill-switch mandate, financial incentives for whistleblowers, and workforce-impact reporting amounts to a more prescriptive framework than anything currently moving through Congress. Coxon himself acknowledged the limits: he said the kill switch "may help in the short term but is not sufficient to ensure safety in the long term," and called for "some kind of speed control" in frontier model development3.
The subpoena threat is the structural detail that distinguishes this hearing from voluntary industry appearances elsewhere. Three of the four labs agreed to send representatives only after Menin's office warned they could be compelled, demonstrating that a city council can extract sworn cooperation from frontier AI companies.
The hearing also exposed a jurisdictional tension. An AlleyWatch analysis posed the question directly: "should New York City decide which AI products can be sold and used in New York City?" while acknowledging that "some of the package is reasonable, including incident reporting and protections for employees who raise safety concerns". New York State Assemblymember Alex Bores, author of the state's frontier AI safety law, has separately discussed how states should approach regulation. ANALYSIS The city, state, and federal layers are now all active on frontier AI oversight, with no clear delineation of authority.
What's next
The Council's ten-bill package now enters the committee markup process. Governor Kathy Hochul has called on Washington to act against catastrophic AI risk. OpenAI and Anthropic have both said they would embed independent safety evaluators in their companies and give them employee-like access. Whether those voluntary commitments satisfy Menin's office or accelerate the legislative timeline is the next variable. Coxon's closing assessment sets the bar: "The 'move fast and break things' approach may work for a photo-sharing app, but it does not work for building the most powerful technology in history".