California has enacted a sweeping package of child-safety laws targeting social media and AI chatbots, moving restrictions on algorithmic feeds, autoplay and emotionally sensitive chatbot interactions directly into technology product design.
The Associated Press reported that Governor Gavin Newsom signed legislation allowing penalties of up to $1 million per child when large social-media companies are found negligent in causing harm.
The measures also restrict addictive platform features for children under 16 and impose stronger obligations on companies operating AI companion chatbots.
Algorithmic feeds and autoplay face new restrictions
California says the legislation prohibits social-media platforms from giving users under 16 access to features including autoplay and algorithmic feeds based on users’ history and profiles.
The restrictions target mechanisms that have become fundamental to consumer-software growth: automatically loading another piece of content and using behavioral data to determine what appears next.
That makes the legislation relevant to product teams well beyond legal departments. Engagement mechanics that were historically optimized through A/B testing can now become regulated functionality when minors are involved.
California also said the package strengthens privacy protections, including safeguards involving targeted advertising and student data.
AI companion products must add crisis safeguards
The rules go further for conversational AI.
California’s “Adam’s Law,” named after Adam Raine, requires companion-chatbot operators to implement crisis protocols for suicidal ideation, parental controls and notifications when a child disables safety settings.
The state also said the legislation establishes independent child-safety audits and annual risk assessments for companies operating these products.
The Associated Press reported that the package follows growing concern about children forming relationships with AI chatbots and the possibility that conversational systems may reinforce emotional distress instead of recognizing when human intervention is necessary.
California is regulating the engagement layer
For software companies, the most important feature of the package is where regulation is moving.
Governments are no longer focusing only on what platforms remove after harmful content appears. They are increasingly scrutinizing the mechanics that determine how users interact with software in the first place.
Autoplay, personalization, notifications and conversational persistence are product choices. They are also the same features that can maximize engagement.
Newsom said child safety should be baked into design, placing responsibility closer to the way products themselves are built.
That captures the wider shift for builders: safety requirements are moving upstream, from moderation and compliance teams into product architecture.
For platforms serving young users, the question may increasingly be not only whether content is safe—but whether the software itself has been deliberately designed to keep a child engaged for longer than is healthy.
