The OpenAI copyright lawsuit filed by The New York Times has a new voice on OpenAI‘s side: the U.S. government. The Trump administration submitted a 20-page brief this week arguing that training large language models on unlicensed copyrighted material falls within fair use, and that restricting the practice would hurt American competitiveness in artificial intelligence.
The brief lands in the U.S. District Court for the Southern District of New York, where the Times has argued since 2023 that OpenAI illegally used its articles to train ChatGPT and related models without permission or compensation. The government’s filing does not carry the force of a ruling: the case is still being tried, and the brief’s authors hold no jurisdiction over the outcome. Still, an intervention of this kind from a sitting administration adds political weight to arguments the court has to weigh.
What the OpenAI copyright lawsuit brief argues
The brief leans on an executive order President Donald Trump signed last year, which calls for the United States to retain global leadership in artificial intelligence. “Constraining LLM development under a misunderstanding of fair use doctrine would thwart such creative and scientific progress while hindering American prosperity and economic mobility,” the brief says. Loosening those constraints, the filing argues, is necessary for American AI companies to keep pace with rivals abroad.
The filing centers on fair use, the copyright doctrine that permits use of protected material without permission when the result is sufficiently transformative. Whether AI training clears that bar is the question at the heart of dozens of lawsuits filed against companies including OpenAI, Anthropic, and Google since 2023. Publishers, authors, and artists argue their work was scraped without consent. AI companies argue that ingesting text to train a model works more like a person learning from reading than like copying it outright.
How courts have ruled so far
Recent rulings have mostly favored AI companies on the training question itself. Judge William Alsup ordered Anthropic to pay a $1.5 billion settlement to a group of writers last year, but the fine was for using pirated books sourced from illegal shadow libraries, not for training on copyrighted text itself. “Like any reader aspiring to be a writer, Anthropic’s LLMs trained upon works not to race ahead and replicate or supplant them, but to turn a hard corner and create something different,” Alsup wrote, comparing the training process to a person reading a book before writing one of their own.
That distinction, between how material was obtained and how it was used, is likely to shape how the New York Times case unfolds. The Trump administration brief argues courts should treat AI training itself as transformative, separate from any question of how the training data was sourced. If judges adopt that reasoning, it would remove one of publishers’ strongest legal arguments against companies that train models on their work.
What happens next
The Southern District of New York has not set a date for a ruling on the fair use question, and the case is expected to continue moving through discovery and motions for months. Other publishers watching closely include Condé Nast and News Corp, both of which have filed or joined similar suits against AI companies over training data. A ruling favorable to OpenAI would likely shape how those cases are argued, and how aggressively other AI companies license, or decline to license, copyrighted material going forward.