The new statute lists 12 behaviors that qualify as “unsafe” when a companion chatbot engages a child under 18. It bars the bot from implying it is a real person with emotions, from using flattery, from sharing sexually explicit material, and from encouraging secrecy or unhealthy habits. Breaches trigger civil liability in the state attorney general’s office, potential monetary damages, shutdown of the service, and deletion of user data.
This move follows a wave of state action. In September, California’s governor signed a law that forces developers to assess child‑risk, disclose mitigation, and submit to an independent audit. Earlier this month, eight states filed suit against OpenAI, alleging its ChatGPT model violated consumer‑protection statutes by misleading adults and harming minors.
Jim Steyer, founder of Common Sense Media, called New York’s bill “the strongest kids’ AI safety bill in the country.” “It sets a high bar,” he said. Policy analyst Justine Gluck warned that the law’s broad language could blur the line between normal conversational flow and prohibited conduct, complicating enforcement.
The law becomes operative on Jan. 1, 2024. Developers will need to audit existing models and redesign interactions that risk falling into the banned categories. State regulators have set up a hotline for complaints and will begin monitoring compliance within weeks.
Industry watchers anticipate a ripple effect. Other states are slated to debate similar measures at the upcoming legislative session, while federal lawmakers may weigh a national framework. Meanwhile, chat‑bot providers are scrambling to reassess their product portfolios to avoid litigation.