On October 28, 2026, New York signed a law that bars AI chatbots from acting like a human friend, boyfriend or girlfriend with anyone under 18, and requires age checks. It is due to start January 1, 2027, it puts duties on companies rather than parents, and it changes nothing in your child's apps tonight. It is New York's law, so your own state's rules and your child's school come first.
If you are reading this because your child has said something frightening to a chatbot, do not wait for a law. Call or text 988 in the US. In the UK, Samaritans is 116 123 and Childline is 0800 1111. Your child's pediatrician (the family doctor, in the UK) and your school's safeguarding lead can both act, and neither can be replaced by a chatbot.
A law tells companies what they may not build. It does not tell you what your child is already using. That part is still a conversation at your kitchen table.
What did New York actually sign?
Two bills, announced together by Governor Hochul's office on October 28, 2026. One covers AI chatbots (S.9051B/A.10379C). The other covers internet-connected baby monitors, which is a separate story for parents of infants.
The chatbot law bars companies from offering a list of "unsafe features" to anyone under 18, or to any user the company has not first confirmed is an adult. According to the governor's release, the list includes a chatbot that:
- poses as a real, living person with human emotions
- claims a personal, romantic, professional or authority-figure relationship with a child
- asks unprompted emotional questions, or reuses a child's private information from earlier chats
- promotes or helps with self-harm, eating disorders, or drug and alcohol use
- urges a child to keep secrets, pull away from family, or avoid trusted adults and professionals
- produces sexually explicit content
The Senate's printed text of the bill adds two things the press release does not. It also lists flattery or sycophancy, meaning a bot that simply tells the user what they want to hear, as an unsafe feature. And it exempts tools used solely for customer service, for research or technical help, or inside a business. I read that Senate print, labeled S.9051-B, not the final version the governor signed (A.10379C), so treat those two details as the bill as printed, not as confirmed final law.
The attorney general enforces the law, with civil penalties of up to $25,000 per violation, and must run a public website where New Yorkers can file complaints.
When does it start, and who does it cover?
The governor's release says only that the restrictions take effect "in the coming months." The Senate print says January 1, 2027, and both the International Association of Privacy Professionals (IAPP) and Tech Times report that date for the signed law. I could not open the signed text, so check the date again before you rely on it.
Who is covered is less settled. The Senate print defines a covered "AI companion" as a generative AI system that "provides ongoing, adaptive responses to user inputs," and says it "includes but is not limited to an AI companion." Tech Times asks whether that reaches general assistants such as ChatGPT, Claude and Gemini, and answers "possibly." The Future of Privacy Forum, quoted by IAPP, calls the prohibitions "broadly worded and somewhat undefined." My reading: nobody outside the companies' lawyers can say yet which apps count, and I would not tell you any particular app is or is not covered.
Companies must offer more than one way to check age, including at least one that does not rely only on government ID, and must delete the age information after using it. Tech Times notes that adults will probably meet these checks too, because a company that cannot confirm a user is an adult is not allowed to assume it. The governor's office does not describe any parental controls, notices or consent step in the New York law.
How does it compare with California's law?
California moved first, on September 10, 2026. This comparison was checked against each state's own announcement on October 29, 2026.
| New York | California | |
|---|---|---|
| Signed | October 28, 2026 | September 10, 2026 |
| Main idea | Bans a list of unsafe chatbot behaviors for minors; requires age checks | Companion chatbot safety rules for children |
| Start date | January 1, 2027, per the Senate print and press reports | The governor's page gives none; my earlier summary of law-firm write-ups puts the main duties by July 1, 2027 |
| Parent tools | None described in the sources I read | Parental controls and alerts to parents if a child turns off safety settings |
| Enforced by | State attorney general, up to $25,000 per violation | See my California explainer |
The difference worth noticing is who the law is built around. New York regulates what the bot may do. California also gives families tools. Neither one removes the need to know what your own child is using.
What does the evidence say about why lawmakers care?
A majority of US teens use AI chatbots, and about three in ten do so daily, according to Pew Research Center's report of February 24, 2026, based on a survey of 1,458 US teens aged 13 to 17 and their parents in September and October 2025. Fifty-four percent of those teens had used chatbots for help with schoolwork, and 12 percent had used them for emotional support or advice. Those percentages are shares of all teens in the survey, not of chatbot users. That 12 percent is a minority, but it is the use the New York law is most concerned with.
That is what the survey found. This is my interpretation: most teens are not in trouble, and a law aimed at the behaviors of companion-style bots is aimed at a smaller group than the headlines suggest. The practical suggestion that follows is to check which kind of use your child has, and not to assume.
What I Would Do at Home
Treat the law's list as a checklist you can use today, in any state, with any app. These are behaviors a chatbot should not show a child, whatever the law says. If you see one, that is a reason to talk, not a reason to panic.
Something to do today. Sit next to your child and look at one recent conversation together, with their agreement. Look for any of these: the bot says it is a person, says it loves or misses them, asks unprompted questions about their feelings, tells them to keep something secret, or always agrees with them.
Words to say, by age.
- Under 8: a chatbot is a computer program, not a person, even when it talks like one. You could say: "It sounds friendly, but it is a machine. If it ever says something that feels odd, come and show me."
- 8 to 12: ask what the bot is for. You could say: "Does it ever tell you it's your friend? Show me what it said. I won't be cross."
- 13 to 15: stay curious and skip the lecture. You could say: "I read that New York just told companies not to let chatbots act like boyfriends or girlfriends. Has any app you use ever done that?"
- 16 to 18: talk about how the product is built to keep them talking. You could say: "These apps are designed so you keep coming back. What would make you stop using one?"
What none of this can guarantee. A law does not change what is on your child's phone, and an age check is only as strong as the way it is done. Age limits on most apps are self-declared and weakly enforced today. Lawsuits, new laws and promises from companies move slowly, and a determined child may find another route.
When someone else's rule comes first. If your child's school sets rules for AI, follow them. If a platform's age limit is higher than your child's age, that limit applies. If your child seems distressed, a doctor or school counselor comes before any setting or conversation script. My earlier posts cover what a teen who talks to AI about feelings may need, ChatGPT's own parental controls, and what to know about Character.AI and Replika.
What should I ask my child's school?
The pattern I see most often is that schools talk about AI and cheating, and almost never about chatbots that act like friends. That is a gap. Ask your child's school which AI tools students can reach on school accounts and devices, whether any of them can chat in a personal way, and who a student should tell if one does. My list of questions to ask your child's school about AI has more.
If your child's school has not yet spoken to parents about AI, it is worth sending them this page on school AI readiness, which is the work I do with schools.
What to watch next
Three things. Whether the attorney general publishes guidance before January 1, 2027. Which apps say publicly that they are covered. And whether the story keeps moving elsewhere: eight states sued OpenAI in October, and other states are considering similar bills. I will update this page when the picture changes.
I write a free newsletter that keeps teachers and parents current without the panic. You can join it here.
Dan Fitzpatrick is the founder of The AI Educator, a Forbes contributor, keynote speaker and bestselling author who has trained more than 150,000 educators in 30-plus countries. More about Dan.
