Washington backs OpenAI on fair use in New York lawsuits

The U.S. government has filed a statement of interest in federal court in New York to support OpenAI's argument that training models on news articles falls under fair use. The administration cites competition and national security. The New York Times is contesting this and is seeking compensation for the content used, while other similar cases are multiplying around AI models.

A Already Crowded Legal Front: Authors, Publishers, and a $1.5 Billion Settlement

Beyond major media groups, authors, including George R.R. Martin, have accused OpenAI of using their works to train its models. In a separate case, Anthropic reached a settlement with authors for $1.5 billion. A judge also ruled that the use of protected works by Anthropic fell under fair use. Earlier this summer, publishers criticized OpenAI for not disclosing certain aspects of its training method, which the company contested.

The Federal Government's Argument: Competition, Security, and Access

The administration argues that a favorable ruling for the plaintiffs could hinder the development of AI, restrict competition among companies in the sector, and reduce the competitiveness of the United States against foreign developers. The filing mentions a threat to national security if rules made it significantly more difficult for a robust AI industry to flourish in the U.S. The government also highlights perceived benefits of the models, such as helping independent authors generate images without resorting to additional licenses, or directing users to dissenting sources. It believes that an oligopoly of the largest tech companies in LLM training, maintained by licensing barriers, does not serve the public interest.

The Cases Targeted in New York and the Plaintiffs' Grievances

The statement of interest was submitted on Tuesday in several ongoing proceedings before the federal court for the Southern District of New York, including those brought by The New York Times and Ziff Davis. In 2025, Ziff Davis filed a lawsuit accusing OpenAI of copyright violations related to the training and use of its systems. The New York Times, which initiated its proceedings in late 2023, claims that language models have been built from content including millions of its articles. The publisher argues that this use aimed to profit from its investment in journalism and to create competing products.

The New York Times' Response: Charging for Model Training

A spokesperson for The New York Times, Graham James, criticizes the administration for siding with a small number of AI companies valued at trillions of dollars, rather than supporting American creators. He advocates for AI companies to fairly compensate for the content used, in accordance with copyright law. According to him, allowing the appropriation of content without permission or compensation would jeopardize the sustainability of human-created content, on which society and AI systems depend. This response comes as the administration supports the fair use argument put forth by OpenAI.