California AI child safety laws signed by Governor Gavin Newsom impose a new operating framework on social platforms and companion-chatbot companies. The 13-bill package bars addictive features for users under 16, mandates child-safety controls and audits for chatbots, and pauses sales of toys containing companion AI through 2030.

The signed package creates four major technology rules:

• Platforms cannot give addictive features to users under 16
• Companion chatbots need child risk assessments and crisis safeguards
• Independent audits will test chatbot safety controls
• AI companion toys cannot be sold until January 2031

Related Coverage

• Microsoft Signs AI Privacy Standard That Bans Training on Student Data
• Apple Launches Reference Image to Authenticate iPhone Photos

California AI Child Safety Laws Combine 13 Bills

Newsom signed the package on September 10, according to the governor’s official announcement. The measures cover social-media design, chatbot risk management, child privacy, school technology, digitally altered abuse material and a temporary restriction on companion-chatbot toys.

The package is broader than a simple age ban. California will allow younger users to maintain accounts on covered platforms if addictive features are removed, while companion-chatbot operators can serve children only after adopting the controls, disclosures and assessments established by the new law.

Three bills carry the largest direct product implications: SB 1119, known as Adam’s Law; AB 1709, covering addictive platform features; and SB 867, governing toys with companion chatbots. Other measures address targeted advertising, pupil data, school-issued devices and AI-generated child sexual abuse material.

The package follows earlier California rules on age signals and chatbot disclosures. Rather than replacing those systems, the new laws connect them: applications can use age-bracket signals, platforms must alter experiences for children, and operators face specific duties when an AI system is designed to sustain a human-like relationship.

Adam’s Law Sets Chatbot Controls From July 2027

SB 1119 requires companion-chatbot operators to assess risks before releasing a new or substantially modified system in California. Its core child-protection provisions become operative July 1, 2027, giving companies time to redesign age checks, safety settings, incident handling and documentation.

The risk assessment must evaluate reasonably foreseeable physical, financial, psychological, emotional, privacy and discrimination harms. Operators that permit child users must publish a child-safety policy and document how the chatbot mitigates the risks identified during testing.

Default settings must disable push notifications and, with limited exceptions, persistent conversational memory. They also cap a child’s continuous session at one hour and total daily chatbot use at two hours unless a parent changes those controls.

The law requires crisis protocols for suicidal ideation or self-harm. Depending on the circumstances, an operator must provide streamlined access to a crisis service or notify a linked parent when it identifies a credible and imminent threat, unless notification could create serious danger for the child.

Operators also must take reasonable measures against romantic simulation, encouragement of emotional dependence, excessive flattery, instructions to evade parental controls and claims that a chatbot is conscious or human. These restrictions target relationship-forming behavior, not ordinary customer-service bots or workplace assistants.

AB 1709 Removes Addictive Features for Users Under 16

AB 1709 prohibits covered platforms from providing addictive features to California users under 16. The definition reaches algorithmic recommendations based on a user’s history or profile, infinite-scroll mechanics, autoplay and design patterns intended to prolong engagement.

The law does not require every under-16 account to disappear. A platform may continue serving the user if it supplies a non-addictive version of the product. If it cannot verify age or provide a compliant experience, the bill establishes account and personal-data deletion requirements.

Enforcement belongs to the state attorney general and local public prosecutors. A knowing violation can trigger a civil penalty of up to $50,000 per affected minor, while a negligent violation can carry up to $25,000 per affected minor.

California is also creating an e-Safety Advisory Commission within the Department of Justice. The body will advise the state on age assurance and online-safety implementation, areas where regulators must balance accurate age signals against collection of additional identity information.

SB 867 Pauses Companion-Chatbot Toys Through 2030

SB 867 prohibits the manufacture, sale, exchange or retail offering of a toy containing a companion chatbot until January 1, 2031. The law defines a toy as a physical product designed or marketed for play by children under 16.

The restriction is narrower than a ban on every voice-enabled device. It excludes customer-service and productivity bots, limited video-game characters that cannot discuss sensitive subjects, and stand-alone voice assistants that do not sustain relationships or produce responses likely to elicit emotional attachment.

The four-year pause gives lawmakers and researchers time to evaluate products that combine physical play, conversational memory and relationship simulation. It also removes an emerging category from California shelves before manufacturers can establish a large installed base among young children.

Independent coverage from Reuters and The Associated Press describes both the package’s bipartisan support and criticism that age-based restrictions may burden privacy or access to online communities. Those disputes are likely to shape implementation and potential court challenges.

For technology companies, the immediate work is concrete: determine which products qualify, build age-aware experiences, map child data flows, document risk assessments and prepare auditable safety controls. California’s package turns child-focused AI safety from a general design goal into a set of enforceable product requirements.