California Assembly Bill 2013 (AB 2013), known as the Generative Artificial Intelligence: Training Data Transparency Act, was signed into law on September 28, 2024, and is set to take effect on January 1, 2026. The legislation mandates developers of generative artificial intelligence (AI) systems or services to disclose detailed information about the datasets they used to train their models. It appears to use the terms “data” and “datasets” to broadly mean any item (including raw information, structured data, and copyrightable works) used to train the applicable generative AI system or service, as evidenced by the law’s suggestion that data could be “protected by copyright.” US law does not afford copyright protection to data points but rather to creative expressions.
The law marks a broader legislative effort in California to regulate AI. Other notable laws passed in conjunction with AB 2013 include Senate Bill 942, which requires large AI systems to implement watermarking and detection tools for AI-generated content, and AB 3030, which mandates disclaimers for AI-generated patient communications in healthcare settings. Additionally, the California Privacy Protection Agency has debated and refined a detailed set of regulations, which includes a prescriptive section on automated decision-making technology (ADMT). These new laws reinforce California’s role as a leader among states in the AI regulatory, policy, and legislative space, especially as other proposed bills (even if paused for now) are modeled on AB 2013.
Read full post at Goodwin
- Date Published:6.10.2025
- Original Publication: Goodwin