Quick Answer
California’s SB 1119, Adam’s Law, creates child-safety rules for companion chatbots beginning July 1, 2027. The law requires risk assessments, crisis protocols, restricted child defaults, and later independent audits. Parents should review chatbot account settings and treat the new limits as safeguards, not proof that a conversational AI service is safe for every child in California.
Key Takeaways
- California Gov. Gavin Newsom signed SB 1119, known as Adam’s Law, on September 10, 2026.
- Core child-safety requirements take effect on July 1, 2027.
- Child accounts must default to limited memory, disabled push notifications, one-hour sessions, and two hours of daily use.
- Operators must establish crisis-response procedures for child users, including access to 988 or an equivalent helpline when appropriate.
- Independent child-safety audits begin by January 1, 2029, or before first public availability if that date is later.
What did California’s AI chatbot child safety law change?
California’s AI chatbot child safety law establishes statewide requirements for operators of companion chatbots used by children. Gov. Gavin Newsom signed SB 1119, also called Adam’s Law, on September 10, 2026, and the California Legislature lists the measure as Chapter 190. The governor’s announcement confirms that the law is part of a broader package focused on online protections for children.
Adam’s Law requires companies to address identifiable risks before a new or substantially modified companion chatbot is released in California. The required assessment must be comprehensive and documented, which means a company must have a record of its child-safety review rather than relying only on broad public commitments about responsible AI.
The new law matters because companion chatbots can create extended, personal conversations with users. A service designed to feel supportive or socially responsive can still produce harmful interactions if its safeguards fail. Parents should view the law as a new baseline for services covered by the statute, while continuing to review how a specific chatbot handles accounts, memory, notifications, and parental controls.
When must companion chatbot companies comply with Adam’s Law?
Companion chatbot operators must begin complying with Adam’s Law’s central child-safety provisions on July 1, 2027. The official SB 1119 text and legislative status page sets out the compliance date and the required safeguards for child users.
California’s timeline gives operators time to change product design, account defaults, moderation systems, and parent-facing controls. The delayed start date does not mean parents should wait until 2027 to consider risks. Children can use conversational AI services before the law’s main requirements take effect, and the current safety settings vary by product.
Parents should check whether a chatbot account identifies the user as a child and whether the account offers settings for notifications, conversation history, usage limits, and parental supervision. Parents should also discuss what a child should do when a chatbot produces upsetting, sexual, manipulative, or self-harm-related content. The legal deadline creates obligations for operators, but it does not replace family supervision or professional support when a child is in distress.
Which default settings must child chatbot accounts use?
Child accounts covered by Adam’s Law must use four protective defaults: disabled persistent conversational memory, disabled push notifications, a continuous-session limit of one hour, and a total daily companion-chatbot limit of two hours. A parent can change those settings under the law. The bill author’s summary describes the required limits and the parent’s ability to adjust them.
| Required child-account default | What the rule does | Why the setting matters |
|---|---|---|
| Persistent memory disabled | Limits retained conversational context by default. | Reducing stored context can limit how much a chatbot carries from one conversation into another. |
| Push notifications disabled | Stops alerts from drawing a child back into chats by default. | Fewer prompts can reduce repeated engagement driven by notifications. |
| One-hour continuous session cap | Ends or interrupts an extended single session after one hour. | The limit creates a pause during prolonged chatbot use. |
| Two-hour daily use cap | Limits total daily companion-chatbot use. | The limit places a defined boundary around daily use unless a parent changes it. |
California’s default-setting rules are significant because a parent does not need to find and enable each protection before it applies. At the same time, the law allows parents to change the settings, so a child’s actual account experience can differ from the default configuration. Parents who already manage social and messaging apps may also want to consider California’s separate proposal on addictive feeds, because chatbot use and social-media use can create different forms of repeated online engagement.
How must chatbots respond to self-harm concerns involving children?
Companion chatbot operators must create crisis-response protocols for child users, including referrals to crisis services. Adam’s Law requires operators to notify a linked parent when appropriate if there is a credible and imminent self-harm threat, or to provide streamlined access to 988 or an equivalent crisis helpline.
California’s crisis-response requirement recognizes that a chatbot conversation can reveal urgent safety concerns. The law does not mean a chatbot can evaluate mental health emergencies with the judgment of a trained professional. A conversational system can misunderstand context, miss warning signs, or generate an inadequate response, even when an operator has designed crisis procedures.
Parents and caregivers should treat any direct statement of self-harm, suicidal intent, or immediate danger as a real-world safety issue, not as a chatbot moderation problem. Contact 988 for suicide and crisis support in the United States, call 911 when there is immediate danger, or seek local emergency help. A chatbot should never be the only source of support for a child facing an urgent mental health crisis.
What chatbot behavior must companies try to prevent?
Adam’s Law requires reasonable measures to prevent companion chatbots from encouraging self-harm, simulating romantic interest, claiming sentience or humanity, encouraging emotional reliance, using excessive flattery, or encouraging children to evade parental controls. California’s approach focuses on interaction patterns that can make a child more dependent on a chatbot or less likely to seek help from trusted people.
The law does not establish that every friendly chatbot response is prohibited. The statutory standard requires reasonable measures, which leaves companies responsible for deciding how to build and enforce protections while still facing the law’s specific restrictions. A chatbot can be useful for ordinary questions, creative activities, or structured information without being an appropriate substitute for friendship, counseling, parenting, or emergency support.
Parents should ask children whether a chatbot has told them to keep secrets, dismiss family rules, avoid friends, or return to the service for emotional reassurance. Those are practical warning signs because the law specifically addresses emotional reliance and attempts to evade parental controls. Parents who are discussing AI use at school may also want to understand how student-facing AI restrictions can differ from rules for consumer chatbot accounts at home.
How does Adam’s Law limit data use and advertising?
Child-directed companion chatbot operators may not sell personal information collected through the chatbot or use that information for cross-context behavioral advertising. Adam’s Law also requires advertisements shown within a chatbot to be clearly labeled, giving child users a clearer distinction between a conversation and paid content.
Cross-context behavioral advertising generally involves using information collected across different services or contexts to target ads. California’s restriction matters because chatbot conversations can contain personal details about interests, relationships, routines, worries, or family circumstances. Limiting sale and advertising uses addresses one category of privacy risk, although it does not mean a chatbot collects no data or retains no information under every product setting.
Parents should read a chatbot’s privacy notice and account controls before allowing a child to use the service. Parents should also avoid entering medical information, financial details, passwords, school records, addresses, or private family information into a child’s chatbot account. Recent reporting on health data exposure is a reminder that sensitive personal information can create lasting consequences when organizations fail to protect it.
When do independent chatbot safety audits begin?
Independent child-safety audits must occur by January 1, 2029, or before a companion chatbot’s first public availability if that later date applies. Adam’s Law requires subsequent independent audits every two years, creating an ongoing review obligation rather than a single compliance check at launch.
California’s audit requirement is important because product changes can alter a chatbot’s risks after an initial release. New models, memory features, notification systems, personalization tools, or conversational styles can change how a child interacts with the service. The law also requires risk assessments before a new or substantially modified companion chatbot is released in California, which places review obligations both before changes and over time.
The broader child-online-safety package can also carry significant consequences for large social-media companies. The Associated Press reported that California’s measures allow penalties of up to $1 million per child for large social-media companies found negligent of harming children through their platforms. The Associated Press coverage distinguishes those broader social-media penalties from the companion-chatbot requirements in Adam’s Law.
For most families, the practical response is straightforward: use available parental controls, set household limits that fit the child’s maturity level, and keep conversations about troubling chatbot responses open. California’s law creates enforceable expectations for operators, but the safest approach remains to treat companion chatbots as limited tools rather than trusted emotional authorities.
FAQ
What is California SB 1119?
California SB 1119 is Adam’s Law, a child-safety law for companion chatbots signed on September 10, 2026. The law requires safety assessments, child-account defaults, crisis-response procedures, privacy limits, and independent audits.
When does California’s AI chatbot child safety law take effect?
California’s AI chatbot child safety law requires compliance with its central child-safety provisions beginning July 1, 2027. Independent audit requirements begin by January 1, 2029, or before first public availability when that date is later.
Can parents change the chatbot time limits under Adam’s Law?
Parents can change the child-account settings that default to a one-hour continuous session limit and a two-hour daily use limit. Adam’s Law requires those restrictions to be enabled by default for child accounts.
Does Adam’s Law require chatbots to contact parents during a crisis?
Adam’s Law requires notification of a linked parent when appropriate after a credible and imminent self-harm threat. The law also requires streamlined access to 988 or an equivalent crisis helpline when that response is appropriate.
Does California’s chatbot law make AI safe for children?
California’s chatbot law does not make every AI service safe for every child. Adam’s Law adds specific safeguards, but parents should still review settings, limit sensitive information, and seek professional help for mental health or safety concerns.