US government supports OpenAI in NYT copyright training lawsuit
The Trump administration has filed a 20-page brief supporting OpenAI in The New York Times lawsuit over the use of copyrighted material to train large language models. The filing defends the ChatGPT maker’s unlicensed training practices while the case proceeds in the US District Court for the Southern District of New York.
The United States argues that it has a strong interest in maintaining a robust and competitive artificial intelligence sector and in retaining global AI leadership. The brief cites an executive order signed by President Donald Trump last year and warns that an incorrect application of fair-use doctrine could restrain creative and scientific progress, prosperity and economic mobility.
Fair use is central to the dispute
Models behind services such as ChatGPT, Anthropic’s Claude and Google’s Gemini are trained on extremely large collections of published material, including copyrighted books, articles and other media. Publishers, including The New York Times, contend that using their material without permission for training is unlawful.
The legal question is not presented as straightforward. The debate centres on fair use, the copyright-law exception that can permit use of protected works without permission in specified circumstances. In this litigation, a key issue is whether the use of copyrighted works to train an AI model is sufficiently transformative to qualify for that protection.
The procedural and evidence issues in the same dispute remain relevant because OpenAI and NYT dispute over logs and evidence examines the contest over logs and evidence in the copyright case. The new federal filing addresses the broader policy case for LLM development rather than resolving the underlying claims between the newspaper and OpenAI.
The filing is influential, not a judgment
The administration’s intervention is not a ruling, and the brief’s authors do not have jurisdiction over the court hearing the case. Its position could nevertheless carry weight as the court considers how copyright law should apply to AI training.
Recent AI copyright cases have generally been favourable to AI companies on the training question. Judge William Alsup ordered Anthropic to pay a $1.5 billion settlement to writers whose books were used to train its models, but the penalty concerned the use of illegal shadow libraries to obtain pirated books rather than the training itself.
Alsup compared LLM training with a reader learning from works in order to create something different, rather than reproducing or replacing them. That distinction reflects the transformative-use argument now advanced by the United States in support of OpenAI.
Business implication
Businesses procuring or building generative AI should recognise that the legal status of training data remains contested: a government brief may shape the debate, but it does not settle the New York Times case or remove the need to assess copyright exposure in AI use.