Requires developers of generative AI models, by January 2027, to publicly document on their websites detailed information about training datasets, including sources, contents, and any modifications. Exempts models for aircraft operation and national security purposes from this requirement.
Enacts the Artificial Intelligence Training Data Transparency Act in New York's general business law.
Requires developers of generative AI models or services available to New Yorkers, from January 2027 onwards, to post data documentation used for AI training on their websites.
Stipulates that this documentation must include sources and ownership of datasets, purpose-related data descriptions, data point details, and any proprietary or public domain status of datasets.
Mandates disclosure of whether personal, aggregated consumer information, or synthetic data generation was used.
Exempts models related to national airspace operations or those developed solely for national security, military, or defense purposes, provided they are exclusively for federal entities.
Defines key terms, including artificial intelligence, developer, generative AI, substantial modification, and synthetic data generation, to clarify responsibilities and expectations under the act.
Requires transparency to foster understanding and trust in AI applications and models.