California AI Transparency Act (SB 942 as amended by AB 853)

Applies To

Overview

Operative as of 2 August 2026, the California AI Transparency Act sets disclosure and provenance (content origin) duties for covered providers. These are makers of generative AI systems with over 1 million monthly visitors or users that are publicly accessible in California. The law requires covered providers to offer a no-cost detection tool that enables users and third parties to identify AI-generated content. Latent disclosures (hidden labels that machines can read) must be embedded in AI-generated image, video, and audio content so that the AI origin can be identified even after distribution. License agreements governing use of covered AI systems by licensees must require disclosure features to stay intact, with revocation of access within 96 hours of a detected breach. The Act is enforced through civil actions and contract terms, and licensees' only direct duty is to stop using a system once their license is revoked. Providers must also offer users the option to add a visible (manifest) disclosure. Separate duties for hosting platforms and large online platforms start in 2027, and for capture device makers in 2028.

Key Requirements

• •Applies to covered providers: makers of generative AI systems with over 1 million monthly users, publicly accessible in California.
• •Provide a free, publicly accessible AI-detection tool for content made by the covered system.
• •Offer users the option to include a visible (manifest) disclosure in AI-generated content.
• •Embed latent disclosures in AI-generated image, video, and audio content; text is not covered.
• •Require in licenses that disclosure capabilities stay intact, and revoke access within 96 hours if a licensee disables them.
• •Operative from 2 August 2026, as amended by AB 853.

What Your Organization Must Do

• →Check whether your generative AI system exceeds 1 million monthly users and is publicly accessible in California.
• →Plan a free, publicly accessible detection tool that lets anyone identify content your system generated.
• →Build latent disclosures into AI-generated images, video and audio, and offer users a visible disclosure option.
• →Add license terms requiring disclosure features to stay intact, with access revoked within 96 hours of a breach.
• →Review vendor contracts so your teams know access may be cut if disclosures are disabled.
• →Set an internal readiness date before 2 August 2026, and track the 2027 and 2028 duties for platforms and devices.

Playbook Guidance

Step-by-step implementation guidance for compliance teams.

Frequently Asked Questions

Which companies must comply with the California AI Transparency Act and does it apply to AI deployers or only developers?

Covered providers: those who create a generative AI system with over 1 million monthly visitors or users that is publicly accessible in California. Companies that only license someone else's system are not covered providers, but must stop using it if their license is revoked.

What is the compliance deadline for the California AI Transparency Act SB 942 as amended by AB 853?

The Act becomes operative on 2 August 2026. Compliance programs, vendor contract updates, and detection tool deployments must all be in place before that date, so organizations should begin operational readiness reviews well in advance.

What exactly does the free AI-detection tool requirement mean for covered providers under CA-AITA?

Covered providers must offer a no-cost detection tool that allows users and third parties to identify AI-generated content produced by the covered system. The tool must be publicly accessible, meaning providers cannot restrict it to paying customers or enterprise subscribers.

What is the 96-hour revocation rule under the California AI Transparency Act and who is responsible for enforcing it?

If a licensee removes or disables required AI disclosure functionality, the covered provider must revoke that licensee's access within 96 hours of discovering the breach. The obligation sits with the provider, but enterprises should verify this term is enforceable in their vendor agreements.

How do latent disclosure requirements under CA-AITA differ from visible watermarking or labeling obligations in other AI regulations?

The Act requires both kinds. Covered providers must embed latent disclosures in AI-generated images, video, and audio, and must offer users the option of a visible (manifest) disclosure. AI-generated text is not covered.

What contract changes must enterprises make to vendor agreements to comply with the California AI Transparency Act?

Covered providers must put disclosure-preservation terms in their licenses. Licensees should expect those terms in vendor contracts and plan for losing access if disclosures are disabled.