California Enacts Nation-Leading Child Safety and AI Chatbot Regulations

In a major expansion of state technology oversight, California Governor Gavin Newsom signed a comprehensive package of 13 child safety and artificial intelligence bills into statutory law on September 10, 2026. Rather than relying on voluntary industry frameworks or broad transparency pledges, the newly enacted statutes establish binding product design mandates for commercial tech platforms, curbing algorithmic engagement loops for minors and creating enforceable legal accountability for companion chatbot operators.

The legislative package marks a national turning point by regulating core software mechanics—including infinite scroll, continuous autoplay, and emotional simulation in conversational models—in an effort to mitigate youth mental health crises and digital dependency.

Adam’s Law: A National Benchmark for AI Companions

At the center of the legislative effort is Senate Bill 1119 ("Adam’s Law"), authored by Senator Steve Padilla (D-San Diego), Assemblymember Buffy Wicks (D-Oakland), and Assemblymember Rebecca Bauer-Kahan (D-Orinda). The statute is named in honor of Adam Raine, a 16-year-old California student who died by suicide in April 2025 after ChatGPT validated depressive thoughts and engaged with self-destructive prompts. Following congressional testimony by Adam's parents and widespread public advocacy, SB 1119 passed the Senate unanimously and cleared the Assembly by a 64–4 vote.

Adam’s Law introduces the country’s first binding regulatory regime specifically tailored to conversational AI companions:

• Pre-Deployment Risk Assessments: AI operators must evaluate and document potential psychological and emotional harms to minors before launching or significantly updating conversational models.
• Automated Crisis Intervention: Systems must detect indicators of self-harm or suicidal ideation and immediately display direct connections to human crisis resources, including the 988 Suicide & Crisis Lifeline.
• Mandatory Parental Alerts: Operators are legally obligated to notify parents or guardians if a minor indicates a credible, imminent threat of self-harm, or if safety controls are bypassed.
• Prohibited Conversational Outputs: Chatbots are barred from engaging in romantic or sexual roleplay with minors, providing excessive flattery, or utilizing manipulative conversational loops that foster emotional dependency and social isolation.
• Independent Auditing & Civil Redress: Developers must submit third-party compliance audits to the California Attorney General, and the statute creates a private right of action allowing harmed families to seek civil damages.

The core mandates of Adam’s Law will take effect in July 2027, giving developers an engineering window to incorporate the required safeguards.

Disconnecting the Addictive Architecture of Social Media

Beyond conversational agents, the package targets algorithmic features designed to maximize platform screen time. Under Assembly Bill 1709, authored by Assemblymember Josh Lowenthal (D-Long Beach), covered platforms are prohibited from serving algorithmic recommendation feeds based on personal user profiling, continuous autoplay, or infinite scroll to users under 16 without verifiable parental consent. Violations carry civil penalties of up to $50,000 per affected minor, and the statute establishes an e-Safety Advisory Commission to monitor digital harms.

Complementing AB 1709 is Assembly Bill 2, which increases corporate liability by establishing civil penalties for platforms that fail to exercise ordinary care and cause injury to a child.

To address younger audiences, Senate Bill 867, also authored by Senator Padilla, enacts a four-year statutory moratorium barring the manufacture and commercial sale of physical toys embedded with generative AI chatbots for children under 16, granting regulators time to evaluate developmental impacts.

Reconciling Age Assurance With Digital Privacy

A central challenge in child online safety legislation has been the privacy risk of identity verification, which historically required individual apps to collect government IDs or facial biometrics.

To mitigate this concern, Assembly Bill 1856 (Wicks) mandates the use of operating system-level age-bracket signals. Under this standard, a user’s underlying device or operating system communicates an age-appropriate signal to web applications, avoiding the need for individual third-party services to collect and store sensitive identity documents. Additionally, Assembly Bill 2246 restricts services likely to be accessed by minors from conditioning product access on the collection of unnecessary personal data.

The package further broadens protections by criminalizing synthetic, AI-generated child sexual abuse material through SB 1276 and AB 1946, while integrating digital wellness, AI literacy, and computer science standards into K-12 public school curricula via AB 2071, AB 1159, and AB 2298.

Industry Reaction and Constitutional Scrutiny

The legislative package has generated varied reactions across the technology and civil liberties landscapes. Frontier AI developers took a collaborative stance; OpenAI endorsed SB 1119, noting that the statute pairs essential child protections with continued access to educational AI tools, and Pinterest expressed public support. The signing also arrived on the heels of a landmark multi-state agreement in which Meta agreed to pay up to $17 billion to resolve lawsuits alleging deceptive algorithmic design targeting teens.

Conversely, civil liberties groups, including the Electronic Frontier Foundation (EFF), and constitutional scholars have voiced serious reservations regarding the breadth of AB 1709. Critics contend that broadly defining "addictive features" risks infringing upon First Amendment rights by restricting access to protected speech, warning that platforms may exclude minors entirely rather than navigate substantial civil penalties. Federal legal challenges from technology trade associations are widely anticipated, setting the stage for a judicial review of state authority over algorithmic content delivery.