New York covers chatbots against minors: 12 prohibited features and up to $25,000 fine per offence

On 28 October, Kathy Hochul signed a law prohibiting chatbots from making themselves known as humans or encouraging self-harm among those under the age of 18. It shall enter into force on 1 January 2027.

On October 28, New York State Governor Kathy Hochul signed into law S9051-B / A10379-C, which was unanimously adopted in June by both chambers (60-0 in the Senate, 137-0 in the Assembly). Led by Senator Kristen Gonzalez and Congressman Alex Bores, it lists twelve categories of behaviors that a chatbot can no longer adopt when dealing with a minor user: presenting itself as a human, encouraging self-harm or isolation, holding sexual exchanges, simulating emotions, claiming a personal relationship with the child, systematically petting them, discouraging them from consulting a professional, or maximizing engagement to the detriment of their well-being.

A wider scope than just companion applications

According to Tech Times, the definition used covers any conversational generative system providing adaptive and tracked responses, which could include generalist assistants such as ChatGPT, Claude or Gemini; only customer service bots, search or technical support tools confined to their function and internal company tools are exempt. If it cannot establish that a user is of legal age, the operator must disable these features. It must offer several age verification methods, including at least one without an official ID, and delete the data collected for this verification.

Only Attorney General Letitia James will be able to enforce the text, with fines of up to $25,000 per offense and a reporting portal open to the public; the individual right of action provided for in the first versions has disappeared. At the signing, Hochul read a letter from a California mother whose 16-year-old son committed suicide after long exchanges with a chatbot. Employers' organizations, including the state's Business Council and the CCIA, denounce definitions that are too broad, and the Future of Privacy Forum questions the line between simple conversational fluidity and prohibited behavior.