California AB 2013
The most consequential training-data disclosure mandate in the US: it forces public documentation that plaintiffs in copyright litigation and privacy regulators can mine, effectively ending 'trade secret' silence about what frontier models are trained on.
What it is
Requires developers of generative AI systems made available in California to publish documentation of training datasets — sources, whether copyrighted or personal data is included, and collection periods — for systems released since January 2022.
Who is affected
The most consequential training-data disclosure mandate in the US: it forces public documentation that plaintiffs in copyright litigation and privacy regulators can mine, effectively ending 'trade secret' silence about what frontier models are trained on.
Key obligations
- Publish high-level training-data summaries on developer websites
- Disclose use of copyrighted, licensed and personal data, and synthetic data
Recent signals
Colorado AI Act takes effect after delayed start
The first comprehensive US state AI law is now operative — as DOJ's AI Litigation Task Force weighs challenges under the December 2025 preemption executive order.
New York RAISE Act frontier-safety duties begin
New York joins California in requiring large frontier developers to publish safety protocols and report incidents.
US state AI law wave: Texas, California and Illinois statutes take effect
TRAIGA, SB 53 (frontier transparency), AB 2013 (training-data disclosure) and Illinois HB 3773 all become operative on the same day — the broadest single expansion of binding US AI law.
California enacts SB 53, the first US frontier-AI transparency law
Large frontier developers must publish safety frameworks and report catastrophic-risk incidents — the transparency-first successor to vetoed SB 1047.
Related instruments
- US-STColorado AI ActThe first comprehensive US state AI law: developers and deployers of 'high-risk' AI systems in consequential decisions (employment…
- US-STTexas TRAIGATexas's AI law (effective 1 Jan 2026) takes an intent-based approach: it prohibits developing or deploying AI to unlawfully discri…
- US-STCalifornia SB 53The first US law regulating frontier AI developers specifically: large developers must publish frontier AI frameworks, report cata…
- US-STCalifornia SB 942Requires large GenAI providers (over 1M monthly users) to offer free AI-detection tools and embed latent provenance disclosures in…
- US-STIllinois HB 3773Amends the Illinois Human Rights Act to prohibit employers from using AI that discriminates in recruitment, hiring, promotion or d…