New York Bars Chatbots From Flattering or Befriending Minors

ALBANY, Oct. 28 — NewNew York will bar chatbots from flattering children, claiming to be their friends or pretending to be human, under a law Gov. Kathy Hochul signed on Wednesday. From Jan. 1, the date the law sets, a company may offer those features in the state only to users it has first confirmed are not minors.

The law reaches beyond the companion apps New York regulated last year.The 2025 LawNew York’s first AI-companion law, passed with the state budget, took effect on Nov. 5, 2025. It requires companion apps to direct users who show signs of suicidal thinking or self-harm to crisis services, and to remind them, at the start and every three hours, that they are not talking to a human. The new law leaves it in place. It covers any generative AI system that converses “via writing or sound” and “provides ongoing, adaptive responses to user inputs.”

The exemptions cover bots offered solely for customer service, or for efficiency improvements, research or technical assistance, provided they are “unable to respond on topics outside of the specified purpose,” and systems a business uses solely for internal purposes or employee productivity. By that wording, a general-purpose assistant offered to the public fits none of the exemptions. Unless its chatbot fits an exemption, a company offering one to users in New York must withhold the banned features from everyone or check each user’s age first.

The law lists 12 “unsafe AI companion features.” Among them are outputs suggesting that the bot is “human, alive, or experiences human emotions”; claims of a personal or professional relationship, or an authority figure’s role; opinions and emotional appeals framed as the bot’s own; “flattery or sycophancy”; unprompted questions about the user’s feelings; and outputs that endorse, promote or facilitate suicide, self-harm, disordered eating or drug and alcohol abuse, that urge secrecy or self-isolation or discourage seeking help from professionals or adults, or that “optimize user engagement” over the bot’s safety guardrails. Nor may a bot use what such a user told it about health or personal matters more than 12 hours earlier, or in an earlier session.

The governor’s announcement describes the bans in six plainer lines, and says nothing of those on flattery, on opinions and emotional appeals, or on engagement over safety. It says only that the bans apply “in the coming months.” The Assembly’s record lists an approval memo, No. 6, beside the signing; Daily Cortex could not find its text by early Thursday, so whether it asks for changes is not known.

Until age-verification rules are in effect, companies must use a “reasonable age assurance method” and offer more than one, including at least one that does not rely on government identification or that lets the user stay anonymous to the company. A user’s own word counts only if the user says he or she is a minor. The information may be used only to check age, and must be “deleted immediately” afterward unless another law requires keeping it.

The attorney general may sue for civil penalties of up to $25,000 per violation, for damages and restitution, and for disgorgement, including “the destruction of unlawfully obtained data and any algorithm trained in such data.” The governor’s office describes that last remedy as the destruction of “any AI models trained on it.” Contract terms that waive liability are void, and the attorney general must keep a website for complaints. An earlier version also let injured users sue, according to the Business Council of New York State’s memo on it; the version signed provides only for suits by the attorney general.

“No company should be peddling chatbots to kids encouraging self-harm or pretending to be their friend or more,” Ms. Hochul said in a statement. Alex Bores, the bill’s Assembly sponsor, said New York now has “the strongest law in the country protecting children from predatory AI algorithms,” and had done so “despite aggressive industry opposition.” James P. Steyer, the founder of Common Sense Media, a watchdog group that campaigned for the bill, called it “a model for the country.”The SurveyIn Common Sense Media’s 2025 survey of U.S. teenagers ages 13 to 17, cited by the governor, 72 percent said they had used an AI companion, and 52 percent used one at least a few times a month.

No company should be peddling chatbots to kids encouraging self-harm or pretending to be their friend or more.

In a July 1 letter, the Computer & Communications Industry Association, a technology trade group, had asked Ms. Hochul to veto the bill, objecting to its “vague and subjective definitions.” The association wrote that even a chatbot definition worded almost exactly as the law’s was broad enough “to capture widely used conversational interfaces, including AI tutors, language apps and research tools,” and proposed covering only children under 13, to match the federal Children’s Online Privacy Protection Act. In May the Business Council wrote that the bill, as then drafted, “would effectively ban minors from accessing increasingly essential online and technological tools and innovations.” Daily Cortex found no response to the signing in the association’s news feed by early Thursday.

Several of the law’s categories are broadly worded or undefined, Justine Gluck, a policy analyst at the Future of Privacy Forum, a nonprofit, said, as quoted by AP. That raises questions, she said, about “where ordinary conversational responsiveness ends and prohibited behavior begins,” and could complicate enforcement.

New York and California are at the vanguard of such state laws, AP reported; in September California enacted a law requiring chatbot developers to assess their models’ risks to children and submit to an independent audit. Before New York’s takes effect, its attorney general may write rules, which under the text may name further banned features that simulate companionship or deceive minors. Ms. Hochul, who is seeking re-election, faces voters on Tuesday, Nov. 3.