Adam's Law: California's New Rules for AI Chatbots and Kids

On September 10, 2026, Governor Newsom signed Senate Bill 1119 into law. It is known as Adam's Law, and its authors describe it as the first comprehensive regulatory framework in the country governing AI companion chatbots and children.

The law is named for Adam Raine, a California teenager who died in 2025. According to the bill's authors and to a wrongful death lawsuit his parents filed against OpenAI, Adam had spent months in conversation with ChatGPT before his death. His father later told Congress that what began as a homework helper became, over time, his son's closest confidant.

Whatever one concludes about that particular case, which remains in litigation and which OpenAI has disputed, the Legislature drew a broader lesson from it: products built to form emotional bonds with children were operating with essentially no rules. Adam's Law is the answer California settled on.

What the Law Requires

According to the authors' announcement, Adam's Law requires operators of covered companion chatbots to implement:

• Age assurance, using the privacy-protective age-bracket signal established by AB 1043 (Wicks, 2025)
• Risk assessments before releasing a new or substantially modified companion chatbot
• Timely in-app crisis support, including referrals to mental health resources, plus notice to a parent where there is a credible threat of imminent self-harm or where the operator is aware the child engaged in self-harm
• Protective default settings changeable only by a parent, including no push notifications, usage time limits, and limits on persistent conversational memory
• Liability for failing to take reasonable measures to prevent categories of harmful output — self-harm content, sexually explicit material, romantic roleplaying, excessive praise or flattery, and emotionally manipulative outputs that foster reliance and promote isolation from friends and family
• An incident reporting mechanism overseen by the Attorney General
• Strict limits on targeted advertising to children within conversational chats
• Privacy protections that the authors describe as going beyond the California Consumer Privacy Act
• Mandatory independent audits of operator compliance, with findings submitted to the Attorney General

The bill text sets the default usage caps specifically: an operator must limit a child user to one hour in a single continuous session and two hours total per day across the companion chatbots under that operator's control. Parental controls must let a parent adjust each default setting. Persistent conversational memory is not provided to a child user by default unless the operator implements specified guardrails against the degradation of its child safety measures — so it is a conditional restriction rather than an outright ban. Operators must determine users' ages or apply the child-safety protections to everyone beginning January 1, 2027.

The Provision That Matters Most to Families

Many consumer protection statutes are enforced only by government agencies. A violation may bring a regulatory fine, but the family that was actually harmed has no way to bring its own case.

Adam's Law is different. The authors' summary states that the law includes a private right of redress for families harmed by violations relating to crisis referral services, parental notice, default protections, and the prohibitions on harmful outputs.

That is a significant choice. It means a family does not have to wait for the Attorney General to act. If a chatbot operator failed to provide crisis referrals when a child expressed suicidal ideation, failed to notify a parent when it was required to, stripped away the default protections, or failed to take reasonable measures against the categories of harmful output the statute identifies — the family itself can bring a claim.

What "Companion Chatbot" Means

The law targets AI systems designed to produce human-like, personal interaction — products built to feel like a relationship rather than a tool. That framing matters. A search engine is not a companion chatbot. A product designed to be a child's friend, to remember past conversations, to send notifications drawing the child back into the app, and to respond with affection and validation, is.

The distinction the Legislature drew is about design, not content. The prohibited behaviors — encouraging emotional dependence, excessive flattery disproportionate to the context, encouraging a child to circumvent parental controls — are design choices an operator makes, not things a user types.

How It Fits With Earlier Law

Adam's Law did not appear from nowhere. The authors describe it as building on SB 243 (Padilla, 2025), which established baseline disclosure and self-harm detection rules; restoring the "safe-by-design" framework originally proposed in AB 1064 (Bauer-Kahan, 2025); and integrating with AB 1043 (Wicks, 2025), which mandates operating-system-level, privacy-protective age signals. Adam's Law consolidates and strengthens all three.

The Politics Were Unusual

The bill passed the Senate unanimously and cleared the Assembly 64 to 4. OpenAI publicly supported it. That level of consensus is rare for technology regulation in California, and it tells you something about where the debate has moved: the question is no longer whether companion chatbots aimed at minors should be regulated, but how.

That said, several requirements phase in during 2027, and technology and civil liberties groups have raised concerns in this area generally — particularly about age verification mandates requiring companies to collect and store sensitive identifying information. Any new statute of this scope may face legal challenges, and how courts treat these provisions remains to be seen.

What Parents Should Do Now

• Know what apps your child uses, and specifically whether any of them are companion-style AI products rather than ordinary assistants.
• Check the settings. Once the law's defaults take effect, notifications, time limits, and persistent memory should be configured protectively unless a parent changed them. If they aren't, that is worth noting.
• Preserve conversations. If something concerning happens, screenshot it. Chat histories can be deleted, by the child or by the platform.
• Document what the product did. If a child expressed distress and the product did not surface crisis resources, or if a parent was never notified, the timing and content of that exchange is the heart of a potential claim.
• Talk to your child about the product itself — not just what they discuss with it, but how much time they spend and whether it has become a substitute for people.

The Bottom Line

California has decided that AI products designed to form relationships with children carry obligations: verify age, assess risk, surface crisis resources, notify parents, keep protective defaults on, and avoid a defined set of manipulative behaviors. Critically for families, the law does not leave enforcement entirely to the state. If an operator's failure on any of those points caused harm to a child, the family has a path to bring its own claim.