New York has enacted one of the nation's most aggressive laws targeting artificial intelligence chatbots that interact with children, as state governments increasingly step into a regulatory vacuum left by Congress.
Gov. Kathy Hochul signed legislation Wednesday that defines a dozen prohibited “unsafe AI companion features” for users under 18. The law, effective Jan. 1, allows the state attorney general to sue developers who violate the rules, with penalties including fines, forced shutdown of the chatbot, and destruction of data.
The move comes amid growing alarm over the rapid spread of AI tools and their potential to harm minors, and as Washington has largely declined to act. President Donald Trump has dismissed concerns about AI risks and warned states against overregulation, but state legislatures are pressing forward with their own consumer protection measures.
New York and California are leading the charge. In September, California Gov. Gavin Newsom signed a law requiring AI developers to assess and mitigate risks to children, including independent audits. That state also banned addictive features for users under 16 on social media platforms.
Eight states have also sued OpenAI, the maker of ChatGPT, accusing the company of violating consumer protection laws by harming children and deceiving adults. Advocacy groups have called it the most significant coordinated state action against an AI firm to date.
Under New York's law, chatbots cannot suggest they are real people with emotions, opinions, or a genuine relationship with a child. They are also barred from using flattery, sexually explicit content, or encouraging secrecy or unhealthy behavior.
Jim Steyer, CEO of Common Sense Media, praised the law as “the strongest kids’ AI safety bill in the country.” Justine Gluck, a policy analyst at the Future of Privacy Forum, said the differing approaches in New York and California could serve as blueprints for other states when legislatures reconvene.
Gluck noted that the law marks a shift from basic disclosure requirements to regulating the core design and operation of chatbots. But she also flagged that some categories are vaguely worded, creating potential confusion about where ordinary conversation ends and prohibited conduct begins, which could complicate enforcement.
As states experiment with early voting trends and other policy fronts, AI regulation is emerging as a new battleground. The outcome of these state efforts could shape how AI is deployed nationwide, especially as legal battles over election integrity and other issues continue to intensify.
