Generative Artificial Intelligence - Training Data Transparency

HB 823 requires developers of generative AI systems (like those creating text, images, or video) to publish detailed documentation about their training data before releasing or significantly updating these systems. Specifically, developers must post on their website by January 1, 2026, and after each major update, including data sources, how the data supports the AI's purpose, and dataset size (static or estimated for dynamic data). The law applies to public-facing AI systems released after January 1, 2022, but excludes specialized uses like hospital medical tools, aircraft safety systems, federal national security applications, and privacy protection tools. This mandate aims to increase transparency about the data powering AI systems that generate content for public use.