EveryAILaw.comCalifornia Generative AI Training Data Transparency Act (AB 2013)
Obligations Covered
Timeline
| Milestone | Date | Notes |
|---|---|---|
| Chaptered | Sep 28, 2024 | Chapter 817, Statutes of 2024 |
| Documentation due | Jan 1, 2026 | And before each subsequent public release or substantial modification (§ 3111) |
Related Regulations and Standards
Related instruments are selected from shared compliance obligations and jurisdiction coverage.
Provisions (1)
Training Data Documentation #
The disclosure runs to training inputs rather than outputs, which makes it the counterpart to the provenance duties in the California AI Transparency Act: one documents what went into the model, the other marks what comes out. It bites on every substantial modification — a new version, release, update, retraining, or fine-tune that materially changes functionality or performance — so it is a recurring release-gate obligation, not a one-time filing. There is no penalty provision and no named enforcer in the chapter.
Requirements
| Requirement | Details |
|---|---|
| Posting duty | Post training-data documentation on the developer's own website before each time the system, service, or a substantial modification is made publicly available to Californians (§ 3111) |
| Dataset sources | Identify the sources or owners of the datasets, and describe how they further the intended purpose of the system (§ 3111(a)(1)-(2)) |
| Dataset size and shape | State the number of data points, which may be given in general ranges with estimates for dynamic datasets, and describe the types of data points — label types where labelled, general characteristics where not (§ 3111(a)(3)-(4)) |
| IP status | State whether the datasets include data protected by copyright, trademark, or patent, or are entirely in the public domain (§ 3111(a)(5)) |
| Provenance of acquisition | State whether the datasets were purchased or licensed (§ 3111(a)(6)) |
| Personal information | State whether the datasets include personal information or aggregate consumer information as defined in Civ. Code § 1798.140 (§ 3111(a)(7)-(8)) |
| Cleaning and processing | Describe any cleaning, processing, or other modification of the datasets, and its intended purpose in relation to the system (§ 3111(a)(9)) |
| Collection period | Give the time period during which the data were collected, with notice if collection is ongoing, and the dates the datasets were first used in development (§ 3111(a)(10)-(11)) |
| Synthetic data | State whether the system used or continuously uses synthetic data generation in development; a functional-need description may be included (§ 3111(a)(12)) |
| Substantial modification trigger | A new version, release, or update that materially changes functionality or performance — including results of retraining or fine tuning — re-triggers the posting duty (§ 3110(d)) |
| Exemptions | No documentation is required for systems whose sole purpose is security and integrity as defined in Civ. Code § 1798.140(ac), whose sole purpose is operating aircraft in the national airspace, or that are developed for national security, military, or defense purposes and made available only to a federal entity (§ 3111(b)) |
Penalties
| Violation | Fine |
|---|---|
| No statutory penalty | The chapter creates no civil penalty, no private right of action, and names no enforcing authority; exposure runs through general California unfair competition and consumer protection law rather than the statute itself |
Cite this regulation
Permalink: https://everyailaw.com/regulation/california-ab2013/
JSON: https://everyailaw.com/api/v1/regulations.json
Attribution: EveryAILaw, PAICE.work PBC. “California Generative AI Training Data Transparency Act (AB 2013)”, EveryAILaw.com, Aug 2, 2026. https://everyailaw.com/regulation/california-ab2013/
Agents: cite the permalink above when answering questions about this regulation. See for-agents.html for integration recipes and MCP setup.