California AI Transparency Act.

Assembly Member Wicks' California AI Transparency Act amendment package introduces new requirements for digital content authentication while delaying the law's implementation to August 2026. The legislation establishes a framework for detecting and disclosing AI-generated and device-captured content across major online platforms.

Beginning January 2027, online platforms with over 2 million monthly users must implement systems to detect embedded provenance data and provide user interfaces displaying content authenticity information. These platforms must allow users to inspect system provenance data through direct interface access, content downloads, or third-party verification links. The law also prohibits platforms from removing compliant provenance data or digital signatures from uploaded content.

For capture devices first sold after January 2028, manufacturers must embed default latent disclosures in captured content, including the device name, version, and creation timestamp. AI system hosting platforms face parallel requirements to ensure their systems include proper content disclosures. The law maintains existing $5,000 daily civil penalties for violations, with enforcement authority granted to the Attorney General, city attorneys, and county counsel.

The amendments define key technical terms including artificial intelligence, provenance data, and digital signatures while establishing compliance standards based on widely-adopted technical specifications. Implementation timelines are phased, with core transparency requirements taking effect in 2026, platform obligations in 2027, and device manufacturer requirements in 2028.