California AI Transparency Act: provenance labels for AI media

The short version

• Since August 2, 2026, generative AI providers with over 1 million monthly users in California must embed a hidden ("latent") disclosure in AI-generated images, video and audio, offer a visible label option, and provide a free AI detection tool.
• From January 1, 2027, large online platforms must show users available provenance data and must not knowingly strip it.
• From January 1, 2028, new cameras and phones sold in California must offer, and by default embed, provenance data in captured content.

What the law does

The California AI Transparency Act (Business and Professions Code §§ 22757 to 22757.6) was created by SB 942 (2024) and expanded by AB 853 (2025).

Covered AI providers must include a latent disclosure in AI images, video and audio conveying, where feasible, the provider's name, the system name and version, the creation time, and a unique ID. The disclosure must be permanent or extraordinarily difficult to remove, as far as technically feasible. Providers must also offer users the option of a visible disclosure, and provide a free public detection tool that shows provenance data without exposing personal data. If a licensee disables the disclosures, the provider must revoke the license within 96 hours.

Large online platforms (more than 2 million monthly users) must, from 2027, detect standards-based provenance data and let users see and inspect it.

GenAI hosting platforms must not knowingly offer systems that don't place the required disclosures, from 2027.

Who it protects

The public in California, including people who appear in or are imitated by AI-generated media.

What it means for you

Creators: You will increasingly be able to check whether an image, video or audio clip claiming to show you came from a covered AI tool.

Platforms and developers: Build or adopt provenance standards, detection tools and user interfaces before the 2027 and 2028 deadlines.

What it doesn't do

• It doesn't cover text.
• It doesn't let individuals sue. The Attorney General, city attorneys and county counsel enforce it, with civil penalties of $5,000 per violation, and each day counts as a separate violation.
• It excludes services that offer only non-user-generated video game, TV, streaming, movie or interactive experiences.
• It doesn't make any AI content illegal. It's about labels and detection.

Status and key dates

• 2024: SB 942 enacted (Stats. 2024, Ch. 291).
• 2025: AB 853 enacted (Stats. 2025, Ch. 674), moving the start date to August 2, 2026 and adding duties for platforms, hosting sites and devices.
• August 2, 2026: Operative for covered AI providers.
• January 1, 2027: Large online platforms and GenAI hosting platforms.
• January 1, 2028: Capture device manufacturers.

Frequently asked questions

Does California require watermarks on AI images? Covered providers must embed a latent provenance disclosure in AI images, video and audio and offer an optional visible label.

Can I check if a video of me was AI-generated? Covered providers must offer a free detection tool that checks whether content came from their system.

What you can do

• Use providers' free detection tools to check suspicious media of you.
• Keep original files of your authentic content so provenance can be compared.

Royall helps you keep a record of what you have and haven't consented to.

Sources

• Cal. Bus. & Prof. Code §§ 22757 to 22757.6: https://leginfo.legislature.ca.gov/faces/codes_displayText.xhtml?division=8.&chapter=25.&lawCode=BPC
• SB 942 (2023–2024): https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202320240SB942
• AB 853 (2025–2026): https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260AB853

Last verified: 2026-09-27

This is general information, not legal advice.