California Generative AI Transparency Requirements - AB 2013

Applies To

Overview

AB 2013 took effect January 1, 2026, and establishes a transparency baseline for generative AI training data in California. Covered developers must post publicly accessible documentation on their websites that describes the datasets used to train their generative AI systems, including information about data sources, data types, and any known limitations or biases. The law targets developers of publicly accessible generative AI systems, meaning those that interact directly with end users in California, rather than purely internal or enterprise-only deployments. Enforcement authority and specific penalty provisions are administered under California's existing consumer protection and business regulation framework. The legislation does not establish a dedicated AI regulator but relies on existing state enforcement mechanisms. AB 2013 is part of a broader package of California AI legislation enacted in 2024 aimed at increasing accountability across the generative AI supply chain.

Key Requirements

• •Developers must publish training data documentation on a publicly accessible webpage by January 1, 2026.
• •Documentation must describe the categories and sources of data used to train the generative AI system.
• •Documentation must identify known limitations, biases, or gaps in the training data.
• •Requirements apply to any generative AI system made publicly accessible to users in California.
• •Disclosures must be maintained and updated to reflect material changes to training data or model versions.
• •Failure to comply may trigger enforcement under California consumer protection statutes, though specific penalty amounts are not enumerated in AB 2013 itself.

What Your Organization Must Do

• →Audit all generative AI products and services offered to the public to determine which are in scope under AB 2013.
• →Inventory training datasets for each covered system, documenting data sources, types, and any known quality or bias issues.
• →Draft and publish a training data disclosure page on the company website before or immediately upon the January 1, 2026, effective date.
• →Establish an internal process to review and update published disclosures whenever training data or model architecture changes materially.
• →Coordinate with product and engineering teams to ensure ongoing model updates trigger a disclosure review workflow.
• →Review vendor and third-party AI system agreements to determine whether your organization is a developer or deployer under the statute, and assign compliance ownership accordingly.