On March 10, 2025, the New York Senate voted 60-1 to pass S 934, which requires warnings for generative AI systems. The bill now moves to the New York Assembly, where it joins its companion bill (A 3411) in the Assembly Science and Technology committee. Below is a background and summary of the current version of the bill.

Background

The bill was introduced on January 8 by Senator Kristen Gonzalez, who also was a sponsor of the New York LOADinG Act, which we previously wrote about. Senator Gonzalez ran a version of the bill last year (S 9450), which also passed the Senate and made it to a third reading in the Assembly before the legislature adjourned. An Assembly companion bill (A 3411) was filed in January and is currently in committee. The New York legislature does not close until June 12 such that there is ample time for the bill to pass the Assembly.

Warning Requirement

The bill has only one requirement, that is, owners, licensees or operators of a generative AI system must “conspicuously display a warning on the system’s user interface that is reasonably calculated to consistently apprise the user that the outputs of the generative artificial intelligence system may be inaccurate and/or inappropriate.”

Generative AI system is defined as “any artificial intelligence system whose primary function is to generate content, which can take the form of code, text, images, and more.”

Artificial intelligence is defined as “a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations, or decisions influencing real or virtual environments.” The bill goes on to state that AI systems “use machine- and human-based inputs to perceive real and virtual environments; abstract such perceptions into models through analysis in an automated manner; and use model inference to formulate opinions for information or action.” The bill also clarifies that the definition includes systems that use machine learning, large language model, natural language processing, and computer vision technologies, including generative AI.

Ultimately, the bill could, for example, require disclosures on consumer chatbots that use generative AI. However, as companies find more ways to deploy generative AI systems, the bill’s scope could be far reaching. In fact, the bill requires disclosures to “users” but does not define that term and, therefore, it cannot be assumed that it only applies in business-to-consumer interactions.

Enforcement

Failure to provide the required disclosure could result in a civil penalty of the lesser of $25 per user of the generative AI system or $100,000. Each calendar year is a separate violation. The bill does not specifically state that it is enforceable only by the state attorney general’s office.

Effective Date

If passed by the Assembly and approved by the governor, the bill would go into effect 90 days after becoming law.