🇺🇸California

Detailed overview

California is one of the most important AI jurisdictions in the United States. It regulates AI through targeted state statutes rather than through one general AI Act. The two most important California AI laws are the California AI Transparency Act, SB 942, and the Transparency in Frontier Artificial Intelligence Act, SB 53.

The California AI Transparency Act applies to covered providers of generative AI systems. A covered provider is a person that creates or produces a generative AI system with more than 1,000,000 monthly visitors or users and that is publicly accessible within California. A generative AI system is AI capable of generating synthetic content, including text, images, video and audio, that emulates the structure and characteristics of its training data. The Act becomes operative on 1 January 2026.

Covered providers must make an AI detection tool available to users at no cost. The tool must allow users to assess whether image, video or audio content was created or altered by the provider's generative AI system. It must output system provenance data, must not output personal provenance data, must be publicly accessible, must allow users to upload content or provide a URL, and must support an API. Covered providers must also avoid unnecessary retention of submitted content and personal information.

The Act also requires content disclosure mechanisms. Covered providers must offer users an option to include a manifest disclosure in AI-generated or AI-altered image, video or audio content. A manifest disclosure is one that is easily perceived and understandable by a natural person. Covered providers must also include a latent disclosure in such content where technically feasible and reasonable. A latent disclosure is not necessarily visible to a person, but it must be detectable by the provider's AI detection tool and must convey provenance information such as the provider name, system name and version, creation or alteration time and a unique identifier.

Violations of the California AI Transparency Act may result in a civil penalty of USD 5,000 per violation, recoverable in a civil action by the Attorney General, a city attorney or a county counsel. Each day of violation is treated as a separate violation.

California's SB 53 regulates frontier AI developers. It creates the Transparency in Frontier Artificial Intelligence Act and applies to large frontier developers of the most advanced AI foundation models. A foundation model is an AI model trained on a broad data set, designed for generality of output and adaptable to many tasks. A frontier model is a foundation model trained with computing power greater than 10²⁶ integer or floating-point operations. A large frontier developer is a frontier developer with annual gross revenues above USD 500 million.

Large frontier developers must write, implement and publish a frontier AI framework. This framework must describe how the developer incorporates standards and best practices, assesses catastrophic-risk thresholds, applies mitigations, reviews deployment decisions, uses third-party assessments, updates the framework, protects unreleased model weights, identifies and responds to critical safety incidents, and manages internal governance. The framework must be reviewed at least annually.

Frontier developers must report critical safety incidents to the Office of Emergency Services within 15 days of discovery. If the incident creates an imminent risk of death or serious physical injury, the developer must disclose the incident within 24 hours to an appropriate authority. Large frontier developers may also need to submit confidential summaries of catastrophic-risk assessments for internal model use.

SB 53 also includes whistleblower protections for covered employees who disclose information about catastrophic risks or violations. Failure by a large frontier developer to publish or transmit required documents, report incidents, comply with its frontier AI framework, or avoid materially false or misleading statements may result in civil penalties up to USD 1,000,000 per violation, enforceable only by the California Attorney General.

Practical requirements & details

Sourced from California SB 942 (the AI Transparency Act) and SB 53 (the Transparency in Frontier Artificial Intelligence Act).

SB 942 — scope

• Applies to covered providers of generative AI systems — persons creating or producing a generative AI system with more than 1,000,000 monthly visitors or users that is publicly accessible within California.
• Generative AI system = AI capable of generating synthetic content (text, images, video, audio) that emulates the structure and characteristics of its training data.
• Operative on 1 January 2026.

SB 942 — AI detection tool

• Free of charge, publicly accessible, supports upload or URL input, exposes an API.
• Outputs system provenance data, but not personal provenance data.
• Must avoid unnecessary retention of submitted content or personal information.

SB 942 — disclosures

• Manifest disclosure — easily perceived and understandable by a natural person; user opt-in for AI-generated or altered image/video/audio.
• Latent disclosure — required where technically feasible and reasonable; detectable by the provider's detection tool; conveys provider name, system name and version, creation/alteration time and a unique identifier.

SB 53 — scope & defined terms

• Foundation model — AI trained on a broad data set, designed for generality of output, adaptable to many tasks.
• Frontier model — foundation model trained with compute > 10²⁶ integer or floating-point operations.
• Large frontier developer — frontier developer with annual gross revenues above USD 500 million.

SB 53 — frontier AI framework duties

• Write, implement and publish a frontier AI framework covering standards, catastrophic-risk thresholds, mitigations, deployment review, third-party assessments, model-weight protection and internal governance.
• Review the framework at least annually.

SB 53 — incident reporting

• Report critical safety incidents to the Office of Emergency Services within 15 days of discovery.
• Imminent risk of death or serious physical injury — disclose within 24 hours to an appropriate authority.
• Confidential summaries of catastrophic-risk assessments may be required for internal model use.

Whistleblower protections (SB 53)

• Cover employees disclosing information about catastrophic risks or violations.

Penalties

• SB 942: civil penalty of USD 5,000 per violation, recoverable by the Attorney General, a city attorney or a county counsel; each day of violation = separate violation.
• SB 53: civil penalties up to USD 1,000,000 per violation for failure to publish required documents, comply with the frontier AI framework or avoid materially false/misleading statements — enforceable only by the California Attorney General.

Related entries

See also the federal-level entry for the United States, which covers the broader US AI regulatory framework (executive orders, NIST AI Risk Management Framework, sectoral regulators, and US-wide federal initiatives).

United States — federal AI framework