Implement Child Safety Requirements for AI Products

This statute requires developers of AI products and chatbots to implement specific safety guardrails to protect minors from harmful content, including self-harm guidance, sexual material, and predatory interaction patterns. It mandates age verification and transparency regarding how data from children is collected, stored, and used for model training purposes. The initiative would apply to any AI product or service with more than one million monthly active users in California, covering a broad range of consumer-facing chatbots, image generators, and recommendation systems. Covered companies would be required to conduct and publish annual Child Safety Impact Assessments detailing the risks their products pose to users under 18 and the mitigation measures in place. Age verification requirements under the statute allow for multiple compliance methods, including device-level age signals, third-party verification services, and self-declaration with parental confirmation for users under 13. The proposal creates a private right of action allowing parents and guardians to sue AI developers for actual damages and statutory damages of up to $5,000 per violation when safety guardrails fail and a minor is exposed to harmful content. Thomas Hiltachk, a prominent California political attorney who has managed numerous statewide ballot campaigns, filed the initiative with the Attorney General’s office and must collect approximately 547,000 valid signatures to qualify for the November 2026 ballot. Technology industry groups including TechNet and the Chamber of Progress have expressed opposition, arguing that the statute’s broad definitions of harmful content could require overly aggressive content filtering that degrades product quality for all users. Supporters including Common Sense Media, the California PTA, and several child psychology research organizations argue that voluntary industry commitments have proven insufficient and that enforceable standards are necessary. The statute also requires AI developers to provide parents with tools to monitor and restrict their children’s interactions with AI products, including conversation logs and usage time limits. An enforcement division within the California Attorney General’s office would be authorized to investigate complaints, conduct audits of covered companies, and impose civil penalties of up to $25,000 per day for noncompliance. If enacted, the statute would take effect on January 1, 2027, with a 180-day grace period for companies to achieve compliance with the technical requirements.