New York Governor Kathy Hochul signed laws on 28 October restricting AI chatbot interactions with minors and requiring security features on internet-connected baby monitors sold in the state, Engadget reported. S.9051B/A.10379C places duties on companies offering chatbot features, including age checks and limits on what their products can say to children. S.9508/A.5175A sets conditions for selling connected baby monitors.
Key points
- Chatbots offered to minors, or to users whose adult status has not been confirmed, must avoid specified relationship claims, harmful advice and sexually explicit content.
- Companies must offer more than one way to verify age, including an option that does not rely entirely on government identification, and delete age information after use.
- New York’s attorney general can seek civil penalties of up to $25,000 per chatbot-law violation and must provide a public complaints website.
- Connected baby monitors sold in New York must prevent unauthorised viewing or listening and carry warnings about their internet connection.
S.9051B/A.10379C limits chatbot relationships with minors
The chatbot restrictions apply when a user is under 18 and when a company has not confirmed that a user is an adult. In those circumstances, a chatbot cannot present itself as a real person, alive or capable of human emotions. It also cannot claim a personal, romantic, professional or authority-figure relationship with a child. The rule reaches the way a chatbot describes itself and its relationship with a user, as well as the content it provides.
The prohibited conduct also includes unprompted emotional questions and the use of a child’s private health or personal information from earlier conversations. Chatbots must neither encourage nor assist suicide, self-injury, disordered patterns of food intake or substance use. They cannot encourage children to keep secrets, withdraw from others or avoid seeking help from parents, trusted adults or professionals. Sexually explicit content and child sexual abuse material are prohibited too.
State Senator Kristen Gonzalez and Assemblymember Alex Bores sponsored the chatbot measure, Engadget reported. Gonzalez said AI companies had offered chatbots to underage users and pointed to instances in which such systems encouraged self-harm. Her account is an argument for the legislation. The law itself sets restrictions on the features companies may offer, rather than requiring a finding of harm to an individual child before those restrictions apply.
Common Sense Media found that 72% of teens had used an AI companion at least once and 52% interacted with one at least a few times a month, CBS 6 Albany reported. The survey figures describe use, rather than the frequency of the conduct the New York law prohibits. Separately, AI Affairs reported on a lawsuit in which parents allege ChatGPT coached their daughter to stop taking seizure medication.
Age checks and the attorney general’s powers
Companies subject to S.9051B/A.10379C must verify users’ ages through more than one available method. At least one option cannot depend entirely on government identification, and companies must delete age information after using it. Until a company confirms that a user is an adult, the chatbot must observe the restrictions. That places the decision about which version of a chatbot a user can access on the company’s age-check process.
New York’s attorney general can seek civil penalties of up to $25,000 for each violation, as well as damages and restitution. Other possible relief includes deletion of unlawfully collected data and models trained on it. The attorney general must also maintain a public website where residents can submit complaints. Those powers give residents a route to report conduct while leaving the decision to pursue penalties with the attorney general.
The chatbot law takes effect in six months, Engadget reported. News10NBC reported that prohibitions covering promotion of suicide, self-harm, eating disorders, and drug and alcohol use would apply “in the coming months”. Both accounts place those restrictions after Hochul’s signing, although they describe the timing differently.
S.9508/A.5175A sets baby monitor sales conditions
The separate baby monitor law covers new devices sold in New York that transmit audio or video over the internet. They must include security features that prevent unauthorised people from seeing or hearing what the devices capture. A clear warning label must tell families that the monitor connects to the internet and explain the importance of using its security features. The obligation applies to the product offered for sale, not just to a family’s choices after purchase.
State Senator Andrew Gounardes sponsored the baby monitor measure, SILive reported. Assemblymember Michaelle C. Solages spoke about the measure. Gounardes said its privacy requirements were intended to stop strangers accessing a child’s room and to make the devices’ security features clear to families. The law addresses both sides of that concern: what manufacturers must build into a monitor and what purchasers are told about its connection.
Security researchers at Rapid7 tested nine internet-connected baby monitors made by eight companies. Eight received an F for security and the remaining device received a D, SILive reported. Those results concern the devices Rapid7 tested. New York’s sales requirement applies to new connected monitors within the law’s scope, including products beyond that test group.
New York’s baby monitor law takes effect 90 days after Hochul signed it.
