Law / United States / 
California
AI Training Data Transparency Act (AB 2013)
Cal. Civ. Code §§ 3110-3111 (AB 2013, Ch. 817, Statutes of 2024)
In force since , under challenge in court.
A crawl signals rule binding public and private bodies.
- How the hook was established
- express
- What makes it apply
- Market targeting
- Audit expectation
- none
- Court action
A suit challenging this law is pending. No court has blocked it. Premium data
What it requires
- If you develop a generative AI system made available to Californians, publicly post documentation of your training datasets before each release or substantial modification, starting .NIST AI RMFMAP 2.3
- State whether those datasets include copyrighted, trademarked, or patented data, and whether they include personal information.NIST AI RMFMAP 4.1
These lines are LexLint's reading of the law's text.
In the frameworks
- NIST AI RMF
- MAP 2.3Scientific integrity and TEVV considerations are identified and documented, including those related to experimental design, data collection and selection (e.g., availability, representativeness, suitability), system trustworthiness, and construct validation.
- MAP 4.1Approaches for mapping AI technology and legal risks of its components – including the use of third-party data or software – are in place, followed, and documented, as are risks of infringement of a third party’s intellectual property or other rights.
Our reading of where this law's duties sit in each framework, and in the NIST AI RMF, line by line beside each duty above. It never says that running a control meets the law. How we read a framework against the law
Who enforces it
Settledness
A Ninth Circuit appeal, No. 26-1591, from the Central District of California, No. 2:25-cv-12295, was pending with an amicus brief filed on .
- As of
- Under challenge
- Yes
- Open questions
- How specific must "the sources or owners of the datasets" be within the "high-level summary of the datasets" under Civil Code section 3111(a)(1), when a developer collected data by crawling the web?
- How does a developer that scraped web data determine "whether the datasets include any data protected by copyright, trademark, or patent" under Civil Code section 3111(a)(5)?
What it makes you log
Who may demand the log
Public
What the log must hold
Data lineage
Logging duty
Section 3111(a) requires a developer, on or before and before each release or substantial modification of a generative artificial intelligence system or service made available to Californians, to post documentation of its training data: the sources or owners of the datasets, the number and types of data points, whether they include copyrighted material or personal information, any cleaning or processing, the collection period and the dates the datasets were first used. The duty is a published disclosure that cannot be written without a record of the data used; the section sets no log format, retention period or audit. Subdivision (b) exempts systems whose sole purpose is security and integrity, the operation of aircraft in the national airspace, or national security use by a federal entity. The row cites section 3111; the definitions in section 3110 are not coded.
- Kind
- Implicit
- As of
- Provision
- Cal. Civ. Code § 3111(a)
- Trigger
- content_generation
What this law does
AB 2013 requires developers of generative AI systems made available to Californians, released or substantially modified since , to publicly post documentation of their training datasets, including whether the datasets include data protected by copyright, trademark, or patent, and whether they include personal information as defined in Civ. Code § 1798.140. The duty commences and repeats before each subsequent release or substantial modification.
AB 2013 added Title 15.2 (commencing with Section 3110) to Part 4 of Division 3 of the Civil Code. It was chaptered as Chapter 817, Statutes of 2024. A federal district court denied a preliminary injunction against enforcement in X.AI LLC v. Bonta (C.D. Cal., ), finding no likelihood of success on trade-secret or constitutional claims without reaching the law's ultimate validity, so the law remains in effect subject to that ongoing litigation.
When LexLint raises it
When your app profile says your app trains models or crawls the web.
Back to the example · Lint your app
Page updated .