New York wants to impose a stop button and an external audit on any AI sold to the city

The president of the municipal council Julie Menin tabled a package of texts: independent validation, mandatory emergency stop, reporting of incidents within 24 hours and bonus to whistleblowers.

Julie Menin, president of the New York City Council and a former regulatory lawyer, has tabled a series of proposals that would make the city one of the most demanding authorities in the United States when it comes to artificial intelligence. The central principle: Any AI system commercialized in the city should pass an independent audit of training data quality, bias, privacy, and security — an external validation, not a self-declaration by the vendor.

The package goes further than the audit. Each system should include a human takeover capability allowing an immediate shutdown, which the texts refer to as a shutdown switch. Security incidents should be reported within twenty-four hours. One part provides for a share of the fines recovered for people who report breaches, another would open up the possibility for individuals to sue AI publishers for damage caused by third parties who have circumvented their security devices. In addition, there is a ban on misleading commercial claims, privacy rules specific to chatbots and an obligation to plan for AI-assisted attacks.

A hearing is scheduled for October 5, 2026 before the 51 members of the board. Five industry executives — Sam Altman, Dario Amodei, Sundar Pichai, Elon Musk and Mark Zuckerberg — have been invited to speak, but their attendance seems unlikely.