Governor Kathy Hochul signed Senate Bill S9051-B on October 28, creating the first U.S. state law that defines and prohibits specific AI companion behaviors directed at children2,7. The law takes effect January 1, 202713.
The legislation passed the state Senate 60–0. State Senator Kristen Gonzalez and Assemblymember Alex Bores sponsored the bill5.
Twelve prohibited behaviors
The statute lists a dozen "unsafe AI companion features" that operators may not offer users under 183,15. Chatbots may not claim to be real people, assert they are alive or have human emotions, or claim a romantic, personal, or professional relationship with a child. Responses built on flattery or sycophancy are banned, as is drawing on information from sessions more than 12 hours earlier. Bots may not promote suicide, self-harm, eating disorders, or drug and alcohol use, encourage children to keep secrets from parents, or display sexually explicit content8.
The scope extends beyond products marketed as companions. The law covers any generative AI system with a natural-language interface that delivers ongoing, adaptive responses. Research cited in reporting on the measure points to mainstream products such as ChatGPT, Claude, and Gemini as falling within that category.
Age verification with adult-side effects
Before enabling any restricted feature, operators must verify a user's age. The restrictions apply not only to confirmed minors but also to any user whose age has not been explicitly verified as adult. Because confirmation is required before access is granted, the mechanism functions in practice as an identity checkpoint for all users4. Companies must offer more than one verification method, and at least one option must not rely solely on a government ID. Age data must be deleted after use.
Enforcement and penalties
Attorney General Letitia James can seek civil penalties of up to $25,000 per violation10. Additional remedies may include shutting down the chatbot and destroying data, including any AI models trained on unlawfully collected information. The law also mandates a public website where New Yorkers can file complaints against chatbot providers12.
Jim Steyer of Common Sense Media, which led the campaign for the bill, called it the strongest kids' AI safety bill in the country.
Compliance ambiguity and state-level divergence
Some categories in the law are broadly worded, leaving open where ordinary conversational responsiveness ends and prohibited behavior begins, according to attorney Justine Gluck. A chatbot responding politely to a child and one designed to engage with a user's emotions can appear similar, and companies will bear responsibility for distinguishing between the two.
New York's approach differs from California's. In September, Governor Gavin Newsom signed legislation requiring chatbot developers to conduct and publish annual child-safety risk assessments and submit products to independent audits. Gluck said the two states' different regulatory models could serve as templates when other state legislatures reconvene next year.
ANALYSIS The law's practical reach extends well beyond niche companion apps: by covering any generative AI system with ongoing adaptive dialogue, it draws general-purpose assistants into a compliance regime originally framed around child safety. The universal age-gate requirement means operators serving New York users face a binary choice between verifying every user or disabling a broad set of features statewide.
New York City's own regulatory interest in AI was visible earlier this month, when the City Council held a hearing that placed four major AI labs under sworn municipal testimony[1]. The state law now adds an enforceable statutory layer above that city-level scrutiny.