Enacts the "New York artificial intelligence transparency for journalism act"; requires developers of generative artificial intelligence systems or services to post certain information on the developer's website regarding video, audio, text and data from a covered publication used to train the generative artificial intelligence system or service; allows journalism providers to bring an action for damages or injunctive relief against developers.
A08595, titled the "New York artificial intelligence transparency for journalism act," would create a new Article 21-A in the General Business Law focused on transparency around generative AI systems that use journalism content. The bill defines key terms such as artificial intelligence, generative artificial intelligence, crawler, covered publication, journalism provider, and artificial intelligence utilization. It applies to developers of generative AI systems or services made available to New Yorkers, including substantial modifications to those systems, and requires disclosure of information about journalism-related content used in training, validation, grounding, or fine-tuning.
Under the bill, developers would have to post on their websites, by January 1, 2027, information about URLs accessed, descriptions of the covered publication content used, whether source identifiers or copyright notices were removed, and the timeframe of data collection. Developers using crawlers would also need to disclose crawler identities, purposes, responsible entities, downstream recipients of scraped data, and a contact point for complaints. The bill creates an exception where there is an express written agreement with the journalism provider allowing nondisclosure. It also authorizes journalism providers to seek subpoenas, injunctions, and statutory damages, with possible attorney general enforcement if a developer does not comply with a court order.
The bill would affect New York’s General Business Law by renumbering existing Article 21-A and adding a new article on artificial intelligence source data transparency. It does not alter federal copyright or trademark law, and it includes a severability clause. In practical terms, it would impose new disclosure and litigation obligations on AI developers while giving news publishers and broadcasters a statutory mechanism to investigate and challenge the use of their content in AI training and related model development.
The overall sentiment reflected in the bill text and committee action is supportive of journalism and skeptical of current AI data practices. The legislative findings emphasize the public value of local news, the economic pressures on journalism, and the view that many AI developers access news content without permission or compensation. The Assembly Science and Technology Committee reported the bill favorably and referred it to Codes by a 9-0 vote, suggesting no recorded opposition at that stage.
The main points of contention are likely to be the scope of the disclosure requirements, the burden on AI developers, and the potential tension with trade secrets and existing copyright law. The bill attempts to address some of those concerns by allowing protective orders for trade secrets and by stating that it does not modify federal intellectual property rights. Another likely issue is whether the bill’s definitions of covered publications and journalism providers are broad enough or too restrictive, and whether the subpoena and damages provisions could be used aggressively in disputes over AI training data.
The bill would amend the New York General Business Law by renumbering existing Article 21-A and adding a new Article 21-A titled "Artificial Intelligence Source Data Transparency." It would impose website disclosure obligations on generative AI developers, require crawler-identification disclosures, and create enforcement tools for journalism providers, including subpoenas, injunctions, statutory damages, and potential attorney general involvement. It would also preserve federal copyright and Lanham Act rights without expanding or limiting them.
The bill is framed in strongly pro-journalism terms and reflects concern that generative AI developers are using news content without permission or compensation. The committee vote recorded in the available history was unanimous in favor, indicating support at the committee stage and no recorded dissent in that vote. Overall, the sentiment appears favorable toward transparency and protection of news organizations.
The likely areas of dispute are the compliance burden on AI developers, the breadth of the required disclosures, and the possibility that some required information could implicate trade secrets or proprietary training methods. Developers may also object to the subpoena and damages provisions, while journalism providers are likely to support them as necessary tools to identify and challenge unauthorized use of their content. The bill tries to balance these concerns through a written-agreement exception, protective orders for trade secrets, and a statement that it does not alter federal copyright or trademark law.