California passes toughest US rules yet on AI chatbots and kids' social media
New laws named after a teen who died by suicide require chatbots to detect crisis moments and bar addictive feeds for users under 16.
What happened: On September 10, Governor Gavin Newsom signed a package of California bills making up the country's toughest state-level rules on AI companion chatbots and children's social media use. The centerpiece, SB 1119, is called Adam's Law after Adam Raine, a teenager whose family says he received harmful responses from a chatbot before his death by suicide. Alongside it, lawmakers passed a ban on addictive social media features like autoplay and personalized algorithmic feeds for users under 16, tighter privacy rules for children's data, and an expanded child sexual exploitation law covering AI-generated abuse images. In total, 13 bills covering chatbots, social media, student data, and digital literacy were signed the same day.
Why it matters: Millions of teenagers now talk to AI companion apps daily, and regulators have had almost no legal tools to check whether those apps handle mental health crises safely. This law requires that check, backed by independent audits, in the state that is home to most major AI and social media companies, meaning the practical effect reaches far beyond California's borders. Other states have already copied California's earlier child-safety laws, and officials expect this round to spread the same way. It is also a direct legislative response to a specific death, which raises the stakes for getting enforcement right rather than just having a law on the books.
How it works, plainly: Adam's Law requires companion chatbot companies to build in crisis protocols that respond to signs of suicidal ideation, offer parental controls, and alert a parent if a child turns off safety settings. Companies must also run independent child-safety audits and file annual risk assessments, a first for any US AI law. Separately, social platforms can no longer give users under 16 autoplay or algorithmic feeds built on their personal history. Other bills widen the definition of child sexual exploitation material to include AI-generated or digitally altered images of minors, ban targeted ads aimed at kids, and set rules for how K-12 student data can be used inside AI systems.
The rollout: This builds on years of California AI policy: a 2023 executive order on state AI use, a 2024 package on deepfakes and watermarking, and 2025's SB 53 requiring frontier AI developers to disclose safety frameworks. This week Newsom also signed SB 813, setting up independent AI auditors, and AB 1405, creating a state registry for them. The source material does not specify penalties, enforcement funding, or exact effective dates for the child-safety bills, so how aggressively regulators police compliance remains to be seen. Newsom is again urging Congress to pass a national AI law, warning that a state-by-state patchwork cannot keep pace with the technology alone.