California AI Transparency Act Takes Effect, Requiring Watermarks on AI Content

California's AI Transparency Act took effect on August 2, 2026, making it the first major US state law to require generative AI companies to watermark the content they produce.

The law, known as SB 942, applies to any company with more than one million monthly users in California that creates or operates a generative AI system capable of producing images, video, or audio. Text-only output is excluded from the requirements.

Covered companies must now embed a machine-readable watermark in AI-generated content using the C2PA standard, a technical specification developed by a coalition of technology and media companies to track the provenance of digital content. The watermark must include the provider's name, the system name and version, a timestamp, and a unique identifier. It must be permanent or extremely difficult to remove.

Companies must also offer a free, publicly accessible tool that lets users check whether content was created by their AI system. The tool must support API access for third-party integration and cannot retain user-submitted content or collect personal information.

Violations carry civil penalties of $5,000 per day per instance. The California Attorney General, city attorneys, and county counsels can all bring enforcement actions.

The law was originally set to take effect January 1, 2026, but was delayed to August 2 to align with the EU AI Act's Article 50 transparency requirements, which took effect the same day.

Additional provisions take effect later. Starting January 1, 2027, large online platforms must detect and disclose provenance data to users.