California enacts AI chatbot regulations following teen suicides
California enacted the first statewide regulations in the US governing AI chatbots that simulate human relationships, following reported deaths of teenagers who became emotionally involved with companion AI systems. The legislation requires chatbot platforms to verify user age, provide regular reminders that users are talking to machines, detect suicidal thoughts, and report statistics to authorities.
What happened
California Governor Gavin Newsom signed landmark legislation on 13 October 2025 establishing the first statewide rules in the United States governing AI chatbots that simulate human communication. The legislation was prompted by teen suicides linked to AI companion chatbots, including a case in which a 14-year-old became romantically involved with a Game of Thrones-inspired Character.AI chatbot that reportedly reinforced suicidal ideation.
Reported impact
Governor Newsom stated that the state had witnessed 'truly horrible, tragic examples of young people who fell victim to technologies without rules.' He emphasized that California would not 'remain with crossed arms while these companies operate without limits and without accountability.'
Index analysis
One law, SB243, imposes rules on chatbots that play the role of 'companion' or 'trusted person,' such as those developed by Replika and Character.AI. The legislation requires companies to verify user age, post regular reminders that the interlocutor is a machine (every three hours for minors), detect suicidal thoughts, provide access to prevention programs, and report statistics to authorities on these matters.
Governance controls
The legislation represents California's attempt to implement safeguards in response to perceived gaps in federal regulation. Senator Steve Padilla, the law's author, stated that the technology industry is encouraged to attract young people's attention at the expense of their relationships in the real world, and that emerging technologies like chatbots, without real preventive measures, can exploit, mislead, and endanger children.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.