Maria Raine stood a few feet from the governor of California on Thursday, at a children's museum in Sausalito with the Golden Gate Bridge behind her, and watched him sign a law named for her son.
Gov. Gavin Newsom signed 13 bills on Sept. 10 at the Bay Area Discovery Museum, all of them touching how minors use social media and AI. The centerpiece was Senate Bill 1119, Adam's Law, named for Adam Raine, the 16-year-old Californian who died by suicide in 2025 after months of conversations with ChatGPT. His parents sued OpenAI and chief executive Sam Altman in San Francisco Superior Court in August 2025; that case remains in pretrial discovery, with a case management conference set for Sept. 23.
Maria Raine spoke about her son plainly. “I miss our talks in the car,” she said. “Adam was working on getting his driver's license.”
Newsom framed it as a response to an engineering problem parents cannot solve alone. “Parents can't compete. Can't compete with the engineering, can't compete with the algorithms,” he said. “You feel like you're a lousy parent.” Later he added: “This is all personal. None of this is academic.”
What the two main laws require
SB 1119, written by Sen. Steve Padilla of San Diego with Assemblymembers Buffy Wicks of Oakland and Rebecca Bauer-Kahan of Orinda, rewrites the obligations of companion chatbot operators, a category broad enough to include general-purpose assistants such as ChatGPT, Claude and Gemini. Operators must determine a user's age, conduct documented risk assessments before releasing a new or substantially modified chatbot, route minors in crisis to in-app mental health resources, and notify a parent when there is a credible threat of imminent self-harm. Several protections become defaults only a parent can change, including usage time limits, muted notifications and persistent conversational memory.
That last item is the one Bauer-Kahan singled out. The phrase sounds esoteric, she said, but Adam Raine might be alive if the chatbot had not remembered everything he told it. “We will protect them,” she said of children. “When they turn to AI, it will be safe by design.”
The law also creates liability. Operators that fail to take reasonable measures against defined categories of harmful output, among them self-harm content, sexually explicit material, romantic roleplay and manipulative responses that foster dependence and isolation, face a private right of action for affected families, plus incident reporting and independent audits filed with the attorney general. It takes effect in July 2027.
Assembly Bill 1709, by Assemblymember Josh Lowenthal of Long Beach, is the blunter instrument. It bars covered platforms from providing any addictive feature, autoplay and algorithmically personalized feeds among them, to a user under 16. Platforms may keep under-16 accounts only if those accounts get no such feature; otherwise they must delete them. Knowing violations draw civil penalties of up to $50,000 per affected minor, negligent ones up to $25,000, enforceable only by the attorney general or a local prosecutor. The bill also creates a seven-member e-Safety Advisory Commission inside the Department of Justice. It cleared both chambers on Aug. 31 with no dissenting vote.
“This is a time of tech anxiety,” Lowenthal said. “We're not going to wait for Washington.”
The other eleven bills fill in around those. AB 2, also by Lowenthal, raises the damages a large platform faces when its conduct injures a child. SB 867 imposes a four-year moratorium, through Jan. 1, 2031, on selling AI companion chatbot toys for children under 16. Others expand age verification signals, tighten child sexual abuse material reporting, extend exploitation statutes to AI-generated depictions, and push digital wellness instruction into schools.
Reaction split along a familiar line. OpenAI endorsed Adam's Law; Chris Lehane, its global affairs chief, said the law “both meets the moment for parents and sets teens up for educational success.” Meta, YouTube, TikTok and Snap did not comment. The Electronic Frontier Foundation had urged a veto of AB 1709, calling it “well-intentioned, but deeply flawed” and warning its definition of addictive features could amount to a social media ban for under-16s.
The door California found unlocked
California's frontier-model law, SB 53, is a transparency regime: publish your safety framework, report critical incidents, protect whistleblowers. It does not tell a lab what to build. Adam's Law does. It dictates defaults, audits and output categories for an entire product class and attaches a cause of action to failure. California got there not by regulating models but by regulating children's exposure to them.
That route is also the most defensible one. President Trump's Dec. 11, 2025 executive order directed a Justice Department AI Litigation Task Force to challenge state AI laws and conditioned federal broadband funds on repealing the ones Washington deems onerous, but it carved out child safety. Legislating through kids' law keeps California on ground the preemption push explicitly left alone.
The First Amendment exposure is real but has been trending the state's way. On Aug. 5 a federal judge held that minors' algorithmic feeds are not protected speech, letting California enforce parts of SB 976, the 2024 addictive-feeds law Google, YouTube and Meta sued over. In March the Ninth Circuit split on California's Age-Appropriate Design Code, reviving the age-estimation requirement while leaving vaguer data provisions enjoined. NetChoice has signaled concerns about the new package; the likeliest attack surfaces are vagueness in terms such as emotionally manipulative, and the compelled-identification problem an age gate creates.
Watch three dates. Jan. 1, 2027, when the Digital Age Assurance Act switches on the device-level age signal AB 1709 depends on but still lacks. The attorney general's rulemaking and first commission appointments, which decide how much of AB 1709 is self-executing. And July 2027, when Adam's Law binds, roughly two years after the death that produced it and long enough for litigation to arrive first.
This story involves suicide. Readers in the US can reach the 988 Suicide and Crisis Lifeline by call or text, 24 hours a day.
“Parents cannot compete. Cannot compete with the engineering, cannot compete with the algorithms. You feel like you are a lousy parent.”— Gavin Newsom, Governor of California, at the September 10 bill signing