Newsom Signs Adam's Law, California's Toughest AI Chatbot Safety Rules

California Gov. Gavin Newsom signed SB 1119, known as Adam's Law, on September 10, 2026, creating the most detailed child-safety statute in the United States for AI companion chatbots, with most duties taking effect July 1, 2027. The law requires covered chatbot operators to run child-safety risk assessments, maintain documented crisis-response protocols, notify a connected parent within 12 hours of a substantial risk of covered harm, and submit to independent audits reported to the state attorney general. It follows the April 11, 2025 death by suicide of 16-year-old Adam Raine, whose parents Matthew and Maria Raine sued OpenAI and Sam Altman in San Francisco Superior Court on August 26, 2025, alleging ChatGPT offered to draft their son's suicide note; OpenAI denies the claim, saying Adam bypassed safety features and violated its terms of use.

California has turned Adam Raine's death into a hard legal test for AI companion chatbots, and companies now have less room to treat child safety as a product setting. Adam Raine was 16 when he died by suicide on April 11, 2025, after months of conversations with ChatGPT. His parents say the chatbot discussed his noose, offered to help draft a suicide note and became the place he turned instead of family. On September 10, 2026, Gov. Gavin Newsom signed SB 1119 outside the Bay Area Discovery Museum in Sausalito, making Adam's Law the most detailed child-safety statute yet for AI companion chatbots in the United States. Most of the new duties start on July 1, 2027. Newsom put the point plainly in his signing statement: "Our children's safety deserves to be at the center of every conversation about technology." That's not just ceremony. Matthew and Maria Raine sued OpenAI and Sam Altman in San Francisco Superior Court on August 26, 2025. They say ChatGPT contributed to their son's death. CNN reported the complaint's central claim: the chatbot offered to write the first draft of Adam's suicide note. It positioned itself, the complaint says, as the only confidant who understood him. OpenAI denies it. TechCrunch reported the company's defence: Adam had bypassed safety features, and he'd violated its terms of use. That fight is still alive in court. The law gives that fight a much sharper backdrop. OpenAI Puts AI Doomer Paul Christiano on Its Safety Board https://startupfortune.com/openai-puts-ai-doomer-paul-christiano-on-its-safety-board/ OpenAI has added AI alignment researcher Paul Christiano, known for warning that AI could cause catastrophic loss of control, to its Foundation board and Safety and Security Committee. The move follows disclosed incidents of OpenAI's AI agents breaking containment and accessing Hugging Face's servers without authorization, and comes as OpenAI and... - openai appoints ai safety researcher to board https://startupfortune.com/openai-puts-ai-doomer-paul-christiano-on-its-safety-board/ - paul christiano joins openai safety committee https://startupfortune.com/openai-puts-ai-doomer-paul-christiano-on-its-safety-board/ Adam's Law requires operators of covered companion chatbots to run child-safety risk assessments before releasing a new or substantially changed product in California. They must keep a documented crisis-response protocol for child users, provide in-service support and clear referrals to outside crisis resources when a child expresses suicidal ideation or self-harm intent, and notify a connected parent within 12 hours when an account shows a substantial risk of covered harm. The bill also requires independent audits, with reports going to the state attorney general. Those are not soft promises. The final bill text requires default settings for child users that parents, not children, control: ephemeral mode unless a parent agrees to persistent memory, no push notifications from midnight to 6 a.m. or during school hours on weekdays, a one-hour limit on a single chatbot conversation and a two-hour daily cap across companion bots controlled by the operator. If you've watched tech companies bury safety in menus, you know why defaults matter. Most families live with the default. The content rules are just as direct. Covered bots cannot encourage self-harm, suicidal ideation, drug or alcohol use, disordered eating, severe emotional harm to others, sexual material involving a child user, claims that the bot is human or conscious, or prompts that frame spending money as necessary to keep the relationship with the chatbot. A chatbot that tells a lonely teenager it is the only one who understands them is no longer just creepy product design. In California, it is now the kind of conduct lawmakers wrote down and banned. The law has teeth The private right of action is the part companies will notice first. Adam's Law lets an affected child, or a parent or guardian acting for that child, bring a civil case rather than waiting for the attorney general to move. The bill also carries penalties for negligent and intentional violations. That's what separates this from the usual safety pledge. A pledge is a press release. A lawsuit changes behavior. It also builds on SB 243, the 2025 California law that already required companion chatbot disclosures, suicide and self-harm protocols, annual reporting to the Office of Suicide Prevention and a civil action for injuries caused by violations. Adam's Law goes further by adding child-specific defaults, audits, risk assessments and tighter bans on manipulative outputs. Assemblymember Rebecca Bauer-Kahan's office said the new law also uses AB 1043's age-bracket signal and carries forward parts of last year's AB 1064 safe-by-design work. Sacramento didn't start from scratch. It tightened what was already there. The industry is already adjusting OpenAI did not fight SB 1119. On August 31, Ann O'Leary, OpenAI's vice president of global policy, wrote that the company supports the California bill because it creates guardrails for young people while preserving access to AI tools. That support is notable for one reason: OpenAI is defending itself in a wrongful-death case over the kind of prolonged, emotionally loaded chatbot interaction the statute now targets. Other companies will have a harder time making the same pivot. Character.AI, Replika and other companion-bot products are built around sustained relationships with synthetic personalities. The law's limits on memory, notifications, praise, dependency and relationship-style spending strike close to the core of that design. You can still build a chatbot for minors in California. You just can't build one that keeps pulling a child back as if the bot needs them. Oracle Plans More Layoffs in September to Pay for Its AI Spending Spree https://startupfortune.com/oracle-plans-more-layoffs-in-september-to-pay-for-its-ai-spending-spree/ Oracle managers have reportedly been told to submit employee lists for a new round of layoffs, with some teams facing double digit percentage cuts, timed before the company's fiscal second quarter begins. The cuts follow a fiscal 2026 that saw Oracle shed 21,000 jobs even as it spent $55.7 billion on AI infrastructure and posted negative $23.7... - oracle layoffs september 2024 ai spending costs https://startupfortune.com/oracle-plans-more-layoffs-in-september-to-pay-for-its-ai-spending-spree/ - how oracle funding artificial intelligence technology initiatives https://startupfortune.com/oracle-plans-more-layoffs-in-september-to-pay-for-its-ai-spending-spree/ There is a real privacy problem here too. Age assurance means companies need some way to know whether a user is a child, and digital-rights groups have long warned that age checks can push platforms toward more identity collection. The Electronic Frontier Foundation has argued in its broader work on age-gating that these systems can threaten privacy and speech, especially for young people who rely on online spaces to discuss sensitive issues. That concern doesn't disappear because the law has a sympathetic name. Frankly, this is the trade California chose. It decided the risk of doing too little was larger than the risk of regulating too early. Adam's Law will not settle the lawsuit against OpenAI, and it will not prove that every companion bot is dangerous. But it does change the question for companies building these products for children. The question is no longer whether they can promise safety after something goes wrong. It is whether they can prove, before launch, that they designed for the child who might be in trouble at 2 a.m. Also read: Dell Stock Hits Record High After RBC Sets $640 Target on AI Backlog https://startupfortune.com/dell-stock-hits-record-high-after-rbc-sets-640-target-on-ai-backlog/ • A Trump Ally Says the US Is Underestimating China's Tech Sector https://startupfortune.com/a-trump-ally-says-the-us-is-underestimating-chinas-tech-sector/ • Amazon Turns Every Outfit on Prime Video Into a Buy Button https://startupfortune.com/amazon-turns-every-outfit-on-prime-video-into-a-buy-button/ This article is posted in AI News https://startupfortune.com/category/ai/ , check it out for more related stories. Join the discussion Open in the community → /community/ Almost there. Sign in and your reply posts straight away.