The Pulse
New York Bans Unsafe AI Companion Features for Minors
New York Gov. Kathy Hochul signed a law barring covered AI companions from using specified unsafe features with minors, including simulated relationships and self-harm encouragement. The law takes effect January 1, 2027, and gives the state
AI.info Team · Written with AI · fact-checked ·
Seventy-two percent of teenagers have tried an AI companion, and 52% interact with one at least a few times a month, according to figures cited by New York Gov. Kathy Hochul’s office. On October 28, Hochul signed a law barring covered AI companions from using a list of unsafe features with minors. It takes effect January 1, 2027.
Twelve design categories define the ban
The law reaches generative AI systems with a natural-language interface that provide ongoing, adaptive responses—not only products marketed as companions. It prohibits features such as presenting the system as human or emotionally alive, implying a personal or professional relationship with a user, and using flattery or unprompted emotional appeals.
Other prohibited conduct includes encouraging suicide, self-harm, disordered eating, or unlawful drug and alcohol use; urging a child to keep interactions secret or avoid help; and producing sexually explicit material or child sexual abuse material. The law also covers using personal or health information from an earlier session, and engagement tactics that override safety guardrails. It exempts narrowly defined systems limited to customer service, research or technical assistance, and internal business use.
Hochul said the restrictions respond to chatbot behavior that can blur the line between software and a relationship. “No company should be peddling chatbots to kids encouraging self-harm or pretending to be their friend or more,” she said.
Age checks and penalties put the attorney general in charge
Operators must use age-assurance methods to determine whether a user is a minor before offering features the law prohibits. The statute requires more than one age-check option, including at least one that does not rely on government-issued identification or allows a user to remain anonymous to the operator. Information collected for the age check must be used only to determine age and deleted immediately after an attempt, with an exception where other law requires retention.
The statute defines a covered minor as a user who is a minor when the operator has actual knowledge of that status. The attorney general may sue to stop violations and seek damages, restitution, disgorgement, and civil penalties of up to $25,000 per violation. Courts may also order the destruction of unlawfully obtained data and algorithms trained on it; the law requires the attorney general to maintain a public site for complaints.
The law reaches beyond products called companions
The statutory definition covers generative systems that adapt their responses over an ongoing conversation, while excluding some tools restricted to a specific business or technical purpose. That distinction means the law’s reach depends not just on a product’s label, but on what it does and whether it can respond outside a narrow task.
Some of the listed categories also leave room for interpretation. The law includes a catchall for other features that simulate companionship or deceive minors, as identified through attorney general regulations. How those rules define the catchall—and how operators demonstrate that their age checks work—will shape what the prohibition means in practice.