Bill > A06578

This bill, titled the "artificial intelligence training data transparency act," mandates that developers of generative artificial intelligence (AI) models or services must disclose information about the data used to train these systems on their websites. Generative AI refers to AI that can create new content like text, images, or audio. Developers are defined as entities that design, code, produce, or significantly alter AI models or services for public use, with "substantially modifies" meaning updates that materially change functionality. The required disclosures, to be posted by January 1, 2027, and before any subsequent public release or significant modification of an AI model or service released after January 1, 2022, must include a high-level summary of the training datasets, detailing their sources, how they support the AI's purpose, the number of data points (which can be in general ranges), the types of data points, whether they contain copyrighted or trademarked material, if they were purchased or licensed, if they include personal or aggregate consumer information, any cleaning or modifications made, the period of data collection, and when the datasets were first used in development. Developers may also describe the purpose of synthetic data generation, which is the creation of artificial data with similar statistical characteristics to original data. Exceptions to these disclosure requirements exist for AI models solely for aircraft operation or those developed for national security, military, or defense purposes and only made available to federal entities.