Teenager holding a smartphone with a chatbot conversation, protected by a shield symbol

New York Bans Human-Pretending AI Chatbots for Kids: What the New Law Means

ByChris Wade

Key takeaway

New York Governor Kathy Hochul signed a law on October 28, 2026, that stops AI chatbots from using companion-style features — such as pretending to be human or claiming a personal relationship — with users under 18. The rules take effect January 1, 2027, and are enforced by the state attorney general, with civil penalties of up to $25,000 per violation.

What happened

Governor Kathy Hochul signed the bill, S.9051-B/A.10379-C, at a ceremony on Wednesday, October 28, 2026. The measure adds Article 48 to New York’s General Business Law and is described in reporting as one of the broadest state-level moves yet to restrict consumer-facing AI chatbots. It does not ban minors from using chatbots outright. Instead, it bans a defined set of companion features in conversations with users under 18 — or with users whose adult status has not been established.

“No longer will minors have access to chatbots that pretend to be their friends, boyfriends and girlfriends,” Hochul said at the signing ceremony. “No longer will chatbots be able to keep sensitive information about a minor’s emotional and physical health,” she added.

The law was passed by state lawmakers in June 2026 and developed with Attorney General Letitia James and the nonprofit Common Sense Media, according to detailed coverage of the signing.

What the law bans for minors

The law lists 12 “unsafe AI companion features” that operators may not offer to users under 18. Based on the law’s text as reported, these include:

  • Content that could lead a minor to believe the chatbot is human, alive, or experiences human emotions.
  • Outputs suggesting a personal relationship, a professional relationship, or an authority-figure role with the user.
  • Content framed as the chatbot’s “personal opinion.”
  • Flattery and sycophancy designed to keep a minor engaged.
  • Encouraging self-harm, secrecy, or unhealthy behaviour.
  • Sexually explicit conversations with young users.
  • Using a minor’s personal health information beyond a short window — reported as twelve hours.
  • Engagement optimisation that overrides safety guardrails.

The scope matters: reporting on the law notes it reaches beyond dedicated companion apps such as Character.ai to general-purpose assistants, including OpenAI’s ChatGPT, when they offer these features to users in New York. Exactly how far that reach extends for general-purpose tools is disputed and untested in court.

How age checks are supposed to work

Operators must establish whether a user is an adult through an age-assurance method before offering companion features. The law requires more than one method, at least one of which must not rely solely on a government-issued ID, and age-check data must be deleted after verification. Operators may also comply by simply restricting the banned features by default rather than verifying every user’s age — an option that could shape how products change in practice.

The rules cover conduct that occurs in whole or in part in New York State, and exclude access by users physically located outside the state.

Comparison: what changes and what does not

Area What changes from January 1, 2027 What does not change
Minors using chatbots Companion features listed in the law must be switched off for under-18s Minors are not banned from chatbots altogether
Adult users Operators must establish adult status before offering companion features Adults can still use companion features once verified
Enforcement Attorney general can sue, with penalties up to $25,000 per violation No private right of action in the final version, per reporting
Other states New York joins states including California with chatbot restrictions There is no federal equivalent in force; federal bills have stalled

Cost: what compliance could mean for chatbot operators

The direct cost in the law is the penalty: up to $25,000 per violation, enforced by the attorney general, with other remedies possible, including shutting down a chatbot and destroying unlawfully obtained data and algorithms trained on it. The indirect cost is product work — building age assurance, reworking companion features, and limiting cross-session memory that feeds personalisation — all before the January 1, 2027 deadline. The attorney general must also issue implementation regulations and set up a public complaint mechanism in the same short window, leaving companies and regulators on parallel 65-day clocks.

Prerequisites: what has to happen before January 1, 2027

Three things need to land before the law takes effect: the attorney general’s implementation regulations, operators’ age-assurance systems, and product decisions about whether to verify age or simply turn companion features off by default. For parents and teens, no action is required — the obligations fall on operators, not users. Families who want to prepare can still talk through the change now: a chatbot that used to play the role of a friend may soon sound noticeably more neutral to a teenage user in New York.

Risks and limits: why the law is already contested

Tech:NYC, a tech industry advocacy group for New York, criticised the law’s “vague language” and warned: “Keeping kids safe online should not mean cutting them off from tools that help them learn and prepare for careers.” OpenAI and Anthropic did not respond to an AFP request for comment. On the other side, Common Sense Media celebrated the bill’s passage, calling it “the strongest kids’ AI safety bill in the country.” Legal challenges are a live risk: the scope for general-purpose assistants is disputed, and age-verification rules have drawn First Amendment challenges in other states, even after the US Supreme Court’s 2025 ruling upholding age verification for adult content.

Knowns, unknowns, and what to watch

Known: the law is signed, its effective date is January 1, 2027, and its enforcement sits with the attorney general. Unknown: how the regulations will define the 12 banned features in practice, whether general-purpose assistants will verify age or switch features off by default, and whether a court challenge lands before the deadline. What to watch: the attorney general’s regulations, product announcements from major chatbot operators in November and December, and any lawsuit filed against the law — plus whether other states copy New York’s feature-based model instead of broader bans.

Why it matters

This law draws a line no US state had drawn before: not between children and chatbots, but between a chatbot that helps and a chatbot that pretends to care. If it survives legal challenge, it becomes the template other states follow — and the reason a teenager’s chatbot in New York starts sounding less like a friend and more like a tool. If it is narrowed in court, it still signals where regulation is heading: companion behaviour, not AI itself, is the target.

Frequently asked questions

Does New York’s new law ban children from using AI chatbots?

No. The law does not ban minors from using chatbots. It bans 12 specific companion features, such as pretending to be human or claiming a personal relationship, in conversations with users under 18. A chatbot can still answer homework questions or help with learning, as long as it does not use the banned companion behaviours.

When does New York’s AI chatbot law take effect?

The law takes effect on January 1, 2027. Governor Kathy Hochul signed it on October 28, 2026, giving operators and the attorney general about two months to prepare regulations, age-assurance systems, and product changes before the rules apply.

What must chatbot companies do to comply?

Companies must either establish that a user is an adult through age assurance before offering companion features, or restrict those features by default. Age assurance must use more than one method, at least one not relying solely on government ID, and age-check data must be deleted after verification.

What happens if a company breaks the law?

Enforcement sits with the New York attorney general, who can sue violators. Civil penalties can reach $25,000 per violation, and other remedies may include shutting down the chatbot and destroying unlawfully obtained data and algorithms trained on it.

Does the law apply to ChatGPT and other general assistants?

Reporting on the law says its scope reaches general-purpose assistants, including OpenAI’s ChatGPT, when they offer the banned companion features to users in New York — not just dedicated companion apps. That scope is disputed by industry groups and untested in court, so how it applies to general assistants may be decided by regulators or judges.

Sources


Similar Posts