The administration of U.S. President Donald Trump submitted a 20-page brief to the court in support of OpenAI’s position in the lawsuit filed by The New York Times, which argues that using copyrighted material without permission to train large language models is illegal. The brief defends continuing this use as important to preserving the United States’ competitiveness in artificial intelligence.
The brief says that the United States has “a strong interest in continuing to develop a robust and competitive artificial intelligence industry that sets global standards for the practice and procedures of artificial intelligence.” It also links this position to an executive order signed by Trump last year that calls for preserving the United States’ global leadership in artificial intelligence.
The Core Dispute: Is Training Fair Use?
Models such as ChatGPT, Claude, and Gemini rely on massive databases containing published books, articles, and media works. Artificial intelligence companies argue that training models on these works may fall under the “fair use” principle in copyright law, an exception that permits, under certain circumstances, the use of others’ works without a license.
But applying this principle to artificial intelligence training remains the subject of broad legal dispute. The debate focuses particularly on whether using works for training is sufficiently “transformative,” meaning that it produces a function or outputs different from the original works rather than copying or replacing them.
The U.S. administration’s brief warns that restricting the development of large language models because of what it describes as a misunderstanding of the fair use principle could hinder creative and scientific progress and harm prosperity and economic mobility in the United States.
What Does the Government’s Intervention Change?
The brief is not a judicial ruling, and the body that prepared it has no authority to decide the lawsuit. The case is being heard by the U.S. District Court for the Southern District of New York, so the final decision remains with the court. Nevertheless, the administration’s intervention represents an official government position that could add political and legal weight to OpenAI’s arguments, at a time when publishers and authors are trying to define the limits of using their works to train models.
Previous developments in the case indicate that the outcomes of lawsuits involving artificial intelligence training and copyright have, so far, been largely favorable to artificial intelligence companies. Last year, Judge William Alsup ordered Anthropic to pay a $1.5 billion settlement to a group of authors, but the penalty was not connected to the training of artificial intelligence models itself; rather, it concerned the use of illegal shadow libraries to pirate the books used in training.
Alsup likened the training of Anthropic’s models to a person reading a book in order to become a writer, saying that the models were trained on the works not to compete with or replace them, but to depart in a different direction and produce something new.
Why Does This News Matter?
The development confirms that the battle over training data does not concern a single company, but rather a legal path that could determine the ability of artificial intelligence companies to use published content without prior licensing agreements. At the same time, the brief does not resolve the central legal question, nor does it clarify whether the court will accept the government’s interpretation of fair use. The boundaries between legitimate training, the use of pirated sources, and the reproduction of original works remain among the case’s most prominent open questions.