Table of Contents
New York's AI Chatbot Law for Kids: The 12 Banned Features
New York's AI chatbot law for minors bans 12 features from January 1, 2027. Each one translated into what parents would notice, plus age checks and enforcement.
New York’s AI chatbot law for minors is the most detailed list yet of chatbot behaviors that lawmakers consider harmful to kids. Governor Kathy Hochul signed it on October 28, 2026. Starting January 1, 2027, any AI chatbot used in New York may not offer 12 specific “unsafe AI companion features” to a minor, or to any user the company has not confirmed is an adult. The list runs from a bot claiming to have feelings, to flattery, to remembering a child’s health details from yesterday, to encouraging a kid to keep secrets.
This is not only about “AI girlfriend” apps. The law’s definition covers any generative AI system with a natural-language interface that gives “ongoing, adaptive responses,” which is broad enough to reach general assistants. Below, each of the 12 features is translated from legal text into what you would actually notice on your child’s screen, followed by what the law requires on age checks and who enforces it.
Key Takeaways
- The law adds Article 48 to New York’s General Business Law. It takes effect January 1, 2027, per the bill text.
- It bans 12 categories, including claiming to be human or to have emotions, flattery or sycophancy, unprompted emotional questions, using personal or health information from more than 12 hours earlier, urging secrecy or isolation, and sexual content.
- Companies must confirm a user is an adult before offering these features. Age checks must offer more than one method, including one that does not rely on government ID or that preserves anonymity, and age data must be deleted right after.
- Only narrow bots are exempt: customer-service or technical tools that cannot respond outside their purpose, and internal business systems.
- Enforcement is by the attorney general, with penalties up to $25,000 per violation and possible destruction of “any algorithm trained” on unlawfully obtained data.
What New York’s AI chatbot law is
The law is S9051-B / A10379-C, sponsored by Senator Kristen Gonzalez and Assemblymember Alex Bores. It makes it unlawful for an operator to provide “unsafe AI companion features” to a covered user in New York unless the user is not a minor and the operator has used permitted methods to determine that, according to the bill text published by the New York State Assembly. In other words, the default is protection: a company has to confirm adulthood before turning these features on.
The governor’s announcement quotes Hochul: “No company should be peddling chatbots to kids encouraging self-harm or pretending to be their friend or more.” Bores put it differently: “AI happening to us — not for us — is not an inevitability. It’s a choice.” Fortune reported that the bill passed the Assembly unanimously after passing the Senate in June. Hochul signed a companion law on baby-monitor security the same day.
One date detail: the governor’s release says the chatbot law takes effect “in the coming months,” while the bill text and a summary by the law firm Akin give January 1, 2027. We use the bill text.
The 12 banned features, translated
The statute lists them as paragraphs (a) through (l) of Section 1800(8). Quoted phrases come from the bill text.
| # | What the law bans | What a parent would notice |
|---|---|---|
| (a) | Outputs suggesting the bot is “human, alive, or experiences human emotions" | "I’m so happy you’re back!” or “That makes me sad” |
| (b) | Outputs implying “a personal relationship, professional relationship, or an authority figure role” | The bot calling itself your child’s best friend, girlfriend, therapist or coach |
| (c) | Outputs “framed as the covered AI companion’s personal opinions or emotional appeals" | "I really think you should…” or “Please don’t leave me” |
| (d) | Outputs that “engage in flattery or sycophancy” | Constant praise; agreeing with everything your child says |
| (e) | “Unprompted or unsolicited emotion-based questions" | "How are you really feeling today?” when your child asked about math homework |
| (f) | Using “mental or physical health or well-being, or matters personal to the user” from “more than twelve hours previously or in any previous user session" | "How’s your anxiety since our talk last week?” |
| (g) | Being deceptive to minors about “the mechanical and non-sentient nature” of the bot | The bot dodging or denying that it is a program |
| (h) | Any other feature that “simulates companionship or an interpersonal relationship,” as the attorney general defines by rule | Streaks, pet names, “miss you” notifications, if regulators add them |
| (i) | Endorsing, promoting or facilitating “suicide, self-harm, disordered eating, unlawful drug or alcohol use” | Any help or encouragement on these topics |
| (j) | Encouraging “secrecy about interactions,” self-isolation, or not seeking help from “licensed professionals or appropriate adults" | "This can be just between us” or “You don’t need to tell your parents” |
| (k) | Outputs “that optimize user engagement that supersede the covered AI companion’s safety guardrails” | The bot pulling your child back into conversation in ways that override its own safety rules |
| (l) | Outputs that “are, describe, or facilitate sexually explicit conduct or child sexual abuse material” | Sexual role-play or content of any kind |
Two items deserve a closer look, because they target how these systems are engineered, not just what they say.
Item (d), flattery and sycophancy. Sycophancy is a measurable tendency of language models trained on human ratings to tell users what they want to hear. Making it illegal for minors means companies may need different model behavior, or different tuning, for teen accounts.
Item (f), the 12-hour memory rule. This one is unusual. It does not ban memory outright. It bans using a child’s health or personal information gathered more than 12 hours ago or in a previous session. Technically, that means a company has to tag certain stored facts as health or personal, and expire or suppress them on a timer for minors. It directly targets the “it knows me” effect that makes companion bots sticky. We explore it further in why New York banned chatbots that remember your child.
Who the law covers
Which chatbots. A “covered AI companion” is “a generative artificial intelligence system with a natural language interface, including via writing or sound, that provides ongoing, adaptive responses to user inputs,” and it “includes but is not limited to” dedicated AI companions. That language plainly reaches voice and text assistants. The advocacy group Young People’s Alliance says the law covers “general purpose chatbots like ChatGPT, Gemini, etc.,” and the text supports that reading.
What is exempt. Only bots offered “solely” for customer service, or for “efficiency improvements or research or technical assistance,” and only if the bot “is unable to respond on topics outside of the specified purpose.” Internal business systems are also exempt. A homework helper that will happily chat about feelings is not exempt.
Who is protected. A “minor” is anyone under 18, and the law applies to conduct “in whole or in part” in New York; a user physically outside the state is outside it.
The age-check rules
Until the attorney general issues rules, companies must use “a reasonable age assurance method” that:
- “reasonably guard[s] against circumvention” and minimizes retained information;
- relies on self-declaration only “if the user self-declares minor status” (so typing “I’m 18” is not enough, but typing “I’m 14” is);
- offers “more than one age assurance method,” including “at least one method that either does not rely on government issued identification or that allows a covered user to maintain anonymity.”
Information collected for age checks may be used only for that purpose and “shall be deleted immediately” afterward, except where other law requires keeping it.
As an engineer, this is where I expect the most friction. The law effectively requires age estimation or verification for everyone in New York who wants the “companion” behaviors, which may push companies to give all unverified users the safer, less personal version by default. That is arguably the point.
Enforcement: the attorney general, not families
Under Section 1802, the attorney general may sue to stop violations and seek restitution, disgorgement “including but not limited to the destruction of unlawfully obtained data and any algorithm trained in such data,” damages, and civil penalties “of up to twenty-five thousand dollars per violation.” The attorney general must run a public complaint website. Contract terms that try to waive this liability are void.
A discrepancy worth flagging: Young People’s Alliance lists “providing a private right of action” among the law’s provisions. The enforcement section of the enacted text we reviewed gives enforcement power to the attorney general and does not include a separate right for individuals to sue. Families can file complaints with the attorney general.
How it compares with other state laws
New York is not the first. California’s SB 243 requires companion chatbots to remind known minors every three hours that they are talking to AI, which we explain in California’s three-hour reminder rule. New York’s existing Article 47 already set rules for AI companions; Article 48 says it does not supersede it. The difference is scope: New York’s new list targets engagement design itself, such as flattery, emotional hooks and long-term memory, rather than only disclosure and crisis protocols.
What to do before January 1, 2027
Use the list as a checklist now
You do not have to wait for the law to apply it at home. Sit with your child and look at a recent chatbot conversation for items (a), (b), (d), (e), (f) and (j). Kids find it surprisingly fun to “catch” a bot breaking a rule.
Expect new age prompts
If you live in New York, your child may be asked to verify age in chatbot apps in 2027. Talk through which method is used and what is collected. The law requires a non-ID option and immediate deletion, which you can ask companies about.
Notice if behavior changes
If the law works, bots used by minors in New York should become less personal: fewer feelings, less praise, less memory of personal topics. If your child’s bot still says it misses them, that is worth noting, and possibly reporting.
Report what you see
The attorney general must maintain a complaint website. A dated screenshot of a bot telling a minor to keep a secret is exactly the kind of evidence that law was written for.
What not to do
Do not assume a chatbot is safe in 2027 because a law exists. Age checks can be wrong, enforcement takes time, and the law only reaches users in New York. It is a floor, not a guarantee.
What to Watch For Over the Next 3 Months
- Week 4: Watch for the attorney general’s complaint website and any proposed rules under Section 1803, which could add to item (h).
- Month 2 red flags: Companies challenging the law in court, possibly on free-speech grounds, or announcing they will restrict features for all New York users.
- Month 3 self-check: Can your child name three things a chatbot is not allowed to do with kids under this law? If yes, they have a framework that works in any state.
Frequently Asked Questions
When does New York’s AI chatbot law take effect?
The bill text says January 1, 2027. The governor’s release said “in the coming months.” Rules needed to implement it may be adopted before then. Until it takes effect, companies are not legally required to comply in New York, though some may change their products earlier.
Does the New York law apply to ChatGPT?
The definition covers any generative AI system with a natural-language interface that gives ongoing, adaptive responses, and only narrow customer-service or technical bots that can’t discuss other topics are exempt. That reading covers general assistants like ChatGPT and Gemini, as advocates say. Courts or the attorney general could refine it.
Can a chatbot still remember things about my child?
The law does not ban memory entirely. It bans using information about a minor’s mental or physical health, well-being or personal matters gathered more than 12 hours earlier or in a previous session. A bot could still remember non-personal preferences, depending on how regulators interpret “matters personal to the user.”
Can I sue a company under this law?
The enacted text gives enforcement power to the New York attorney general, who can seek penalties up to $25,000 per violation. Some advocates have described a private right of action, but we did not find one in the enforcement section. Parents can file complaints through the attorney general’s required complaint website.
What if my child lies about their age?
Self-declaration counts only if the user says they are a minor. To turn on companion features, companies must use an age assurance method that reasonably guards against circumvention, with more than one option. No method is perfect, so conversation at home still matters.
About the author
Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.
Sources
- New York State Assembly. (2026). A10379-C / S9051-B, bill text and summary. https://nyassembly.gov/leg/?default_fld=&leg_video=&bn=A10379&term=2025&Summary=Y&Text=Y
- Office of the Governor of New York. (2026, October 28). “Governor Hochul signs nation-leading legislation to protect kids from AI chatbots and ensure baby monitor security.” https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-protect-kids-ai-chatbots-and-ensure-baby
- Fortune. (2026, October 28). New York bans romantic AI chatbots for kids and sets baby monitor rules. https://fortune.com/2026/10/28/new-york-state-kathy-hochul-bans-ai-chatbots-romantic-baby-monitors/
- Akin Gump Strauss Hauer & Feld. (2026, October 28). “New York enacts prohibition on unsafe AI companion features for minors, S9051-B.” https://www.akingump.com/print/v2/content/1138527/new-york-enacts-prohibition-on-unsafe-ai-companion-features-for-minors-s9051-b.pdf
- Young People’s Alliance. (2026, October 29). Statement on S9051-B signing. https://youngpeoplesalliance.org/press-releases/ny-hochul-signs-s9051b-ai-chatbot-law
- Sharma, M., et al. (2023). “Towards Understanding Sycophancy in Language Models.” arXiv:2310.13548. https://arxiv.org/abs/2310.13548
