Law / United States / New York / New York City
Third-Party Validation and Shut-Down Capability of Artificial Intelligence ModelsNew, proposed
In committee, dated , as of .
An AI governance rule binding private bodies.
- What it reaches
- supply
As of .
Where it has got to
The text described here is Bill text as posted on the Council's legislation record, with no amendment letter, published .
Locally, this stage is introduced by the Council on and in the Committee of the Whole, which heard it as a pre-considered (P-C) item and laid it over on .
The stage above is recorded at legistar.council.nyc.gov.
The Committee of the Whole heard the item as a pre-considered item and laid it over on . The Council formally introduced it on as Int. No. 1100 of 2026, and the record shows it in the Committee of the Whole. The record carries no amendment letter and does not date the publication of the text, so the version date shown is that of the first recorded action.
What it requires
- This measure is proposed and binds nobody yet; what follows is what it would require if enacted in this form.
- It would make it unlawful to market, offer for sale, sell or deploy an artificial intelligence model that a third-party validator has not validated on task performance, determinism, latency and throughput, data provenance, disparate impact, lawful and secure data handling, and safety, and on any further topics the director of the Office of Cyber Command adds.
- It would make it unlawful to market, offer for sale, sell or deploy an artificial intelligence model that lacks a shut-down capability, the technical capability for a human operator to cause it to stop functioning temporarily or permanently.
- A third-party validator would have to certify to the model's developer and to the Office of Cyber Command whether the model has been validated and is appropriately positioned for deployment, and disclose any financial or other interest the validator has in the model or its developer.
What this law does
This measure is proposed and binds nobody yet.
Sponsored by Council Member Julie Menin, the Speaker, it would make it unlawful to market, offer for sale, sell or deploy an artificial intelligence model unless a third-party validator (a person other than an affiliate of the model's developer, engaged by the developer) has validated it on topics including task performance, data provenance, disparate impact and safety, and the model includes a shut-down capability, meaning the technical capability for a human operator to make it stop functioning temporarily or permanently.
The validator would have to certify to the developer and to the Office of Cyber Command whether the model has been validated and disclose any financial or other interest it has in the model or its developer, and the director of the Office of Cyber Command would write rules on those submissions and on validator qualifications.
The civil penalty would be $25,000 per instance of marketing, offering for sale, selling or deploying a non-compliant model and $25,000 per instance of falsifying a validation, and up to $25,000 for a violation of any other provision of the subchapter, recoverable before the Office of Administrative Trials and Hearings or in a civil action by the corporation counsel, and the bill would take effect 180 days after it becomes law.
When LexLint raises it
When your app profile says your app generates content with AI, deploys a chatbot, makes high-risk automated decisions or distributes a software product.