NYC Council Proposes AI ‘Kill Switch’ as Industry Executives Testify on Safety Risks

The New York City Council convened a rare Committee of the Whole session today, October 5, 2026, to compel testimony from the leadership of the world’s most prominent artificial intelligence laboratories. Executives from Anthropic, OpenAI, Google, and Meta appeared under oath to answer questions regarding the safety, security, and potential existential risks posed by advanced autonomous systems.

The hearing, which involves all 51 members of the Council, was triggered by increasing concerns over the speed of AI development and its impact on public safety. While most invited firms complied with the proceedings, SpaceXAI—the venture led by Elon Musk—failed to appear despite a formal subpoena. Speaker Julie Menin announced during the session that the city will seek judicial enforcement in the State Supreme Court to address the no-show.

The panel of testifying executives included Logan Graham, head of policy at Anthropic; Morgan Dwyer, representing OpenAI; Alice Friend from Google; and Shane Cahill from Meta. According to reports from the hearing, the executives faced sharp questioning regarding their internal “red-teaming” protocols and the degree of control they maintain over frontier models.

Executives from major AI labs faced questioning regarding internal safety protocols and the risk of losing control over frontier models.

The Council is currently evaluating an “AI Package” consisting of 10 separate bills designed to establish local oversight of the industry. The most significant proposal is a mandatory, human-operated “kill switch” for advanced AI models, which would require developers to maintain a physical or manual override capable of shutting down a system if it begins to operate outside of its safety constraints. Other measures in the package include a whistleblower incentive program and a private right of action that would allow individuals to sue for harms caused by AI systems.

The urgency of the legislative push was punctuated by testimony from Jacob Coxon, a former researcher at Anthropic. Coxon told the Council that industry-wide practices are currently insufficient to prevent a “cataclysmic” loss of control. He testified that, in his assessment, humanity is “more likely than not” to lose control of current AI systems, which he claimed could lead to human extinction if development continues without external guardrails.

The legislative scrutiny follows earlier actions by Mayor Zohran Mamdani, who recently implemented a one-year moratorium on generative AI for city students below the ninth grade. That policy took effect ahead of the 2026-2027 school year as officials work to determine the technology’s long-term impact on cognitive development and classroom integrity.

The Council is expected to continue the hearing through the afternoon to review the specifics of the whistleblower protections and the technical feasibility of the proposed kill switch requirements.

The post NYC Council Proposes AI ‘Kill Switch’ as Industry Executives Testify on Safety Risks appeared first on World Weekly News.