California AI Transparency Act.

Assembly Member Wicks' California AI Transparency Act expands the state's oversight of artificial intelligence systems by establishing new requirements for digital content verification and provenance tracking across multiple technology sectors.

The legislation requires large online platforms - those serving over 2 million monthly users - to retain and make accessible provenance data for content posted to their services. This data, which verifies content authenticity and modification history, must either be directly provided to users or indicated through clear notifications. The bill also mandates that manufacturers of cameras, mobile phones, and other capture devices incorporate provenance data capabilities into their default applications and provide secure hardware-based provenance capture access to third-party apps.

These provisions build upon existing requirements for generative AI providers with over 1 million monthly users to offer free detection tools for AI-generated or altered content. Violations incur civil penalties of $5,000 per day, enforceable through actions by the state Attorney General, city attorneys, or county counsel. The legislation authorizes courts to award attorney fees and costs to prevailing plaintiffs in these enforcement actions.

The bill defines several key technical terms, including "artificial intelligence" as engineered systems that generate outputs affecting physical or virtual environments, and "provenance data" as embedded or metadata information verifying digital content's authenticity and modification history. It distinguishes between personal provenance data containing user-identifiable information and system provenance data related to content authenticity and generation methods.