New York City lawmakers subjected technology executives to hours of questioning Monday over the potential dangers of artificial intelligence, as the City Council considers writing its own rules for the rapidly expanding industry.
Council Speaker Julie Menin asked policy officials from Google, Meta, Anthropic and OpenAI to explain how they measure the risks posed by AI in the most extreme, potentially catastrophic scenario. The companies agreed to appear after the council threatened to subpoena them.
Morgan Dwyer, OpenAI’s head of policy development and operations, appeared remotely with the other technology representatives and offered a blunt response: “I don’t know, I also don’t think it matters.”
Menin quickly challenged the answer, calling it dismissive. “To say you don’t know or [it] doesn’t matter is flippant at best,” she said.
She then questioned how OpenAI could determine that its products are safe without being able to define or measure their risks, comparing the approach to a pharmaceutical company releasing a drug while admitting it did not know whether the medicine could kill people.
The marathon hearing began with testimony from AI whistleblowers. Alongside pointed remarks from several council members, their testimony portrayed unrestrained artificial intelligence as a technology capable of destabilizing New York’s economy and critical infrastructure.
Jacob Coxon, a former researcher at Anthropic and OpenAI who warned last month that AI could pose an extinction-level danger, accused the companies of acting recklessly despite the enormous consequences.
“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,” Coxon told the council.
The technology companies offered a more optimistic assessment, emphasizing AI’s potential to accelerate medical discoveries and give small businesses access to sophisticated tools that were once available only to major corporations.
Dwyer said artificial intelligence is already helping address economic, scientific and social challenges at a pace that was previously impossible.
Still, the executives acknowledged that the technology’s most promising breakthroughs come with serious hazards.
Dwyer said AI could deliver enormous benefits but stressed that every technology carries risks, some of which cannot be accepted. His comments came after Menin repeatedly pressed him on the industry’s safety standards.
Logan Graham, head of Anthropic’s Frontier Red Team, which tests advanced systems such as Claude under extreme conditions, agreed that inadequate safeguards could allow AI-related dangers to become “extremely high.”
The executives generally supported the council’s proposal to strengthen AI protections, including the possibility of financial rewards for whistleblowers who disclose serious safety concerns.
Council members also questioned the companies about independent safety reviews, the threat of widespread job displacement and whether they would cooperate with federal immigration authorities.
However, lawmakers did not directly ask the technology representatives about the council’s central proposal: requiring an AI “kill switch” or a human override before such systems could operate in New York City.
The hearing also underscored the limits of the City Council’s authority. While New York officials seek to establish themselves as national leaders on AI regulation, the companies in question operate across state and international borders.
Participants from both government and industry emphasized that meaningful oversight would require strong federal rules capable of keeping pace with rapidly advancing AI systems.
Google’s global director of AI policy, Alice Friends, said the strongest protections would rely on national technical standards applied consistently across state and federal jurisdictions.