Generative Artificial Intelligence: Training Data Transparency (AB 2013)
Requires developers of generative AI systems made available to Californians (released or substantially modified since 1 Jan 2022) to post documentation on their websites about training data, including dataset sources, data types, whether copyrighted or personal information is included, and collection periods. Narrow exemptions cover security, aircraft and federal defence systems.
Why it matters
Every public-facing generative AI provider must maintain a public training-data summary; a federal court refused to block the law in March 2026.
What it requires
Key obligations, provision by provision
2 obligations in article order. How IDs work
Post training-data documentation on your website for each generative AI system or substantial modification released since 1 Jan 2022 and made available to Californians, by 1 Jan 2026 and before each later release.
Include a high-level dataset summary covering sources, purpose, size, data types, copyright or licence status, personal or aggregate consumer information, cleaning, collection periods, first-use dates and synthetic data use.
Cite this record
WTA-R-US-CA-0002: Generative Artificial Intelligence: Training Data Transparency (AB 2013). California Legislature / Governor. wheresthe.ai, https://wheresthe.ai/rule/us-ca-ab2013-training-data-transparency-2024/ (checked 4 Oct 2026).
History
- Effective changed from 1 Jan 2025 to 1 Jan 2026
More from California
-
In forceEffective 1 Jan 2026
-
Partly in forceEffective 1 Jan 2026
-
In forceEffective 1 Jan 2026
-
Enacted, not yet in forceApplies 1 Jan 2027
-
Partly in forceEffective 2 Aug 2026
-
Enacted, not yet in forceApplies 1 Jul 2027