In brief
On October 29, New York Governor Kathy Hochul signed a law banning AI chatbots from offering certain features to users under 18, including pretending to be human and encouraging self‑harm. The legislation caps civil penalties at $25,000, establishes a public complaint website, and follows a trend in other states such as California. The bill’s effectiveness and enforcement remain uncertain.
On October 29, New York Governor Kathy Hochul signed a new law that restricts artificial‑intelligence chatbots from providing specific features to minors, a move described by Xinhua as an effort to protect children from potential harm.
Xinhua reported that the legislation bars companies from offering features that enable a chatbot to "pretend to be human, build emotional dependency, encourage self‑harm or engage in sexually explicit conversations" with users under 18. It also limits the ability of chatbots to pretend to be a child’s friend, boyfriend or girlfriend.
According to the news outlet, violations of the law could trigger civil penalties of up to $25,000, imposed by the state attorney general’s office.
The bill establishes a public website where New Yorkers can file complaints about potential violations, a detail also highlighted by Xinhua as part of the law’s compliance framework.
New York’s action follows a trend of state‑level regulation. Xinhua noted that in September Californian Governor Gavin Newsom signed a requirement for chatbot developers to conduct and publish annual child‑safety risk assessments and to submit their products to independent audits. Newsom’s law also bans social media platforms from deploying addictive features for users under 16.
As the law took effect, Governor Hochul stated:
"No longer will minors have access to chatbots that pretend to be their friends, boyfriends and girlfriends."
— Kathy Hochul, in a statement after the signing ceremony.
In a Washington Post analysis, it was explained that Hochul outlined a dozen definitions of "unsafe AI companion features" that will now be prohibited when a companion chatbot interacts with a child under 18. The article emphasizes that the definitions cover a range of psychological and behavioral risks.
While the legislation sets clear prohibitions and penalties, Xinhua and subsequent reports admit that the law’s enforcement mechanisms remain to be clarified. Questions linger over how the public complaint website will function, who will staff the oversight office, and what criteria will be used to assess violations. Additionally, the California comparison suggests that states may adopt different approaches in measuring child safety risk, prompting uncertainty about inter‑state consistency.
Experts quoted in the Washington Post highlight uncertainty about the law’s practical impact on the rapidly evolving chatbot market. They point out that companies may respond by altering product design or by shifting services outside state borders, strategies that are untested under the new statute.
In sum, New York has enacted a law that restricts several risky chatbot behaviors around children and establishes a penalty framework, but the effectiveness, enforcement, and broader market consequences remain to be seen.
