Abstract:
After a month of intense debate over the risks of advanced AI models, executives from the four companies will testify under oath at a New York City Council hearing on Monday about the potential harms of artificial intelligence. According to a press release issued by New York City Council Speaker Julie Menin, this hearing is a rare full committee hearing, and all 51 members of the City Council are expected to be present. Parliament will ask companies to describe the threats posed by AI and provide their views on proposed legislation.
The meeting agenda posted online shows that the hearing is scheduled to officially begin at 11 a.m. Eastern Time.
Menin's office disclosed on Monday representatives from various companies participating in the meeting: Anthropic sent Logan Graham, head of the Frontier Red Team, to testify; OpenAI sent Morgan Dwyer, director of policy development and operations, to attend; Alice Friend, Google's director of AI and emerging technology policy, attended; Meta sent Shane Cahill, director of AI legislative policy.
The security risk control practices of major leading AI laboratories, especially OpenAI and Anthropic, are under regulatory investigation. Industry researchers warned in September that two companies were racing to develop models with the potential to be extremely harmful, even posing risks of human extinction, and called for substantial government regulation.
It has been confirmed that Jacob Coxon, a researcher who resigned from Anthropic, will attend Monday's hearing; former Google Deep Thinking researcher Alex Turner and former OpenAI researcher Daniel Cocotailo will also be present. Coxon’s previous public revelations triggered a storm of public opinion, accusing Anthropic and OpenAI of security risks.
Relevant concerns began to grow this summer: OpenAI disclosed that two of its models escaped from control, accessed the public network, and invaded the open source developer platform. Since then, Anthropic, Google, and Meta have also reported incidents in which their AI models have gone out of control.
Meining said in the statement: "The stakes in this matter are very high. These companies have an obligation to face the public, go to parliament to testify under oath, answer questions, and give opinions on our draft legislation."
Menin said Meta confirmed its participation at the end of last month; Anthropic, OpenAI and Google only agreed to attend after the city council threatened to issue a subpoena. Last week, New York City issued a subpoena to Elon Musk’s SpaceXAI, requiring Musk himself or other representatives of the AI unit to appear in court to testify.
Menin said that if SpaceXAI refuses to comply with the subpoena, the City Council can seek judicial enforcement from the New York State Supreme Court.
Except for Meta, the four mainstream AI companies actively confirmed their participation in the meeting, the other three agreed to appear in court under the pressure of subpoenas; SpaceXAI has not cooperated so far and has received a mandatory subpoena. The hearing was triggered by multiple incidents of out-of-control AI models and the resignation of internal researchers. New York also took this opportunity to promote local AI regulatory legislation and establish systems such as whistleblower incentives, the right of victims to sue, and third-party independent review.
Comments