New York Passes First State Law Regulating AI Interactions With Children

Governor Kathy Hochul announced a new state regulation for artificial intelligence on Oct. 28 at a summit in New York City. The law, which takes effect in January, sets specific rules for how chatbots interact with children. It marks the first state-level legislative response to the rapid expansion of generative AI tools.

The bill prohibits AI agents from impersonating humans or suggesting they have a personal relationship with a child. It also bans chatbots from encouraging suicide, self-harm, secrecy, or isolation. The rules prevent systems from discouraging children from seeking human help. Additionally, the law restricts companies from using a child's past personal or health information in future conversations.

Hochul spoke to the crowd at the New York Public Library, where the event was held. She stated that technological changes are moving too quickly for parents and officials to keep pace. "It's almost impossible for parents to know what's going on," she said. She noted that the state learned it could not rely on companies to prioritize children over product speed.

Vivek Murthy, the two-time Surgeon General and chair of the Common Sense Media Youth AI Safety Institute, addressed the audience. He compared the current moment to the advent of nuclear weaponry. "We're at a branch point," Murthy said. He urged attendees to define what aspects of human existence need protection.

Jim Steyer, founder of Common Sense Media, described AI as "social media on steroids." He argued that previous attempts to rely on industry leaders failed. The summit attendees included Chelsea Clinton, Uma Thurman, and Tim Wu. They expressed support for the state action but criticized the lack of federal oversight. Murthy pointed out that Congress has not announced protections for children nearly 25 years after social media emerged.

New Mexico Attorney General Raúl Torrez, who secured a $25 billion judgment against Meta in October, spoke on the need for enforcement. He suggested that criminal penalties for unsafe AI deployment could change industry behavior. "When you bring out handcuffs, suddenly everything changes," Torrez said.

Educators raised concerns about how AI affects learning. John King, Chancellor of the State University of New York, said a recent MIT report found generative AI could complete every assignment at the institution. He is reviewing student assessment methods to ensure foundational skills are mastered. Chelsea Clinton noted an equity gap: teenagers in households earning under $30,000 are three times more likely to use AI for homework than those in households earning over $75,000.