Source: Lexology

On September 28, 2024, California Governor Gavin Newsom signed into law Assembly Bill (AB) 2013, which imposes new disclosure requirements on developers of generative artificial intelligence (AI) systems. The law mandates that AI developers publicly disclose detailed information on the datasets used to train their AI systems. The legislation defines AI broadly, covering any system that can generate synthetic content, such as text, images, video, or audio, mimicking the characteristics of the training data.

Developers must post detailed documentation on their websites about the data sources, the purpose of the data, and whether the datasets include copyrighted or personal information. The law also requires developers to disclose whether the data was purchased, licensed, or obtained from the public domain. Compliance is mandatory for all generative AI systems made available to Californians, with a compliance deadline of January 1, 2026, for systems released after January 1, 2022.

The law exempts AI systems designed solely for security, national defense, or national airspace operations. California’s move underscores the importance of transparency in AI development, particularly in ensuring ethical data use and building public trust. Developers may need to adopt new technologies or processes to automate the tracking and disclosure of training data, and those using foundation models should assess how the law impacts their operations.