New York has become the latest and largest US state to regulate artificial intelligence companions, after Governor Kathy Hochul signed a bill restricting how AI chatbots can interact with minors. The measure passed both chambers of the state legislature with unanimous support earlier this year and became the most widely reported AI story of the day, carried by dozens of outlets.
What the AI Chatbots Law Does
The legislation targets the design patterns that keep children talking to chatbots: it restricts features that simulate personal relationships with minors and strengthens requirements around safety disclosures and crisis intervention when a young user expresses distress. Supporters frame it as child protection catching up with technology, noting that companion chatbots are now a mainstream product used by millions of teenagers, often for hours a day and sometimes in place of human confidants.
Part of a wider regulatory wave
New York’s move does not stand alone. In the same week, major AI companies including Google, Microsoft, Anthropic and OpenAI backed a shared technical standard that lets websites state whether their content may be used for AI training, replacing today’s patchwork of crawler rules. Regulators, platforms and publishers are all, in different ways, trying to draw boundaries around systems that scaled faster than the rules written for them. Our blog coverage has followed each step of that negotiation.
Industry reaction and open questions
Enforcement will be the test: age verification is notoriously leaky, and definitions of what counts as a companion chatbot will be argued product by product. Still, unanimous votes are rare in AI policy, and other states are expected to copy the template. A ranked digest of the day’s AI stories, with the chatbot law at the top, is published by this AI news roundup.
The political economy of the law is as notable as its content. Child-safety measures are among the few technology policies that can still command unanimous votes, and chatbot regulation has become the leading edge of a broader settlement in which AI companies accept design duties in exchange for operating legitimacy. Companies will complain, with some justice, that fifty state regimes make compliance a maze; but the same companies have moved quickly to add age signals, parental controls and crisis referrals in the past year, suggesting the direction of travel is accepted even where individual rules are disputed. The next fight is likely to be about verification: how a service proves a user is an adult without building an identity database that creates new risks of its own.
The argument over children and chatbots is really an argument about design responsibility: who answers when a product built for engagement meets a user too young to manage it. New York has given its answer, and the rest of the country is now deciding whether to give the same one.
