California AI Transparency Act

People will have free tools to check whether audio, images, or video were made or materially altered by a provider’s AI. The law aims to make AI-content provenance more reliable while limiting exposure of personal information.

What the law does ​

• Removes the prior 1 million monthly-user threshold, covering any provider of a publicly accessible generative-AI system in California.
• Requires covered providers to offer a free, publicly accessible disclosure-verification tool that checks eligible content or URLs for provider AI creation or alteration and returns detected system provenance data.
• Bars verification tools from disclosing detected personal information unless the affected user gives specific, plain-language consent.
• Requires technically feasible, durable latent disclosures in AI-created or materially altered images, video, and audio identifying the provider, system and version, date and time, unique identifier, and whether AI created or altered the content.
• Requires providers licensing their systems to notify licensees of these duties and, in specified cases, terminate authorization or report unresolved licensee noncompliance to the Attorney General.
• Imposes $5,000 daily civil penalties per violation and a temporary $50,000 daily penalty for falsely claiming a system primarily functions as assistive technology.
• Exempts exclusively non-user-generated videogames and, until January 1, 2029, AI systems primarily designed as assistive technology.

Who it affects ​

• Providers of publicly accessible generative-AI systems in California.
• Third parties licensing and modifying covered generative-AI systems.
• People using AI-generated or AI-altered image, video, and audio content.
• The Attorney General, city attorneys, and county counsel enforcing violations.

Context ​

The urgency law takes effect immediately, while the underlying transparency framework generally begins August 2, 2026.