Justice Department Files Landmark Statement of Interest Arguing AI Training on Copyrighted Works Is Fair Use

Government warns that adverse ruling would threaten national security and hand competitive advantage to foreign rivals

The U.S. Department of Justice has filed a 20-page Statement of Interest in the consolidated OpenAI copyright litigation, marking the federal government’s first direct intervention in the wave of artificial intelligence training copyright cases. The filing urges the U.S. District Court for the Southern District of New York to find that training large language models on copyrighted written works constitutes fair use under Section 107 of the Copyright Act.

The Statement of Interest, filed September 1, 2026, under 28 U.S.C. § 517, advises the court to treat model training — as distinct from the use of AI to generate outputs — as a “highly transformative” fair use under current law. The government explicitly argues that “the training of AI models on copyrighted material, in and of itself, does not violate copyright laws”.

National Security and Economic Competitiveness at Stake

The Justice Department grounded its argument in national security imperatives, citing the Government Accountability Office’s warning that “[f]ailure to adopt and effectively integrate AI technology could hinder national security”. The filing notes that foreign adversaries have similar goals and that rules of law making it “significantly more difficult to develop a robust AI industry in the United States therefore threaten national security and give a competitive advantage to foreign adversaries who are not so encumbered”.

Citing Executive Orders from January 2025 and June 2026, the government emphasized that “[i]t is the policy of the United States to sustain and enhance America’s global AI dominance in order to promote human flourishing, economic competitiveness, and national security”. The brief warned that requiring paid licensing would “put U.S. AI companies at a competitive disadvantage relative to competitors located in other countries that do not consistently respect U.S. intellectual property law”.

Training Is “Exceedingly Transformative,” Government Argues

The government’s fair-use analysis centers on the first statutory factor — the “purpose and character of the use.” The DOJ argues that LLM training is “transformative — spectacularly so,” citing the Northern District of California’s decision in Bartz v. Anthropic PBC. “The purpose of the copying (to build an intelligent, interactive model) differs in kind from the purpose of the copied work (to use language to directly entertain or educate a reading audience)”.

The brief analogizes the current dispute to the Supreme Court’s 2021 decision in Google LLC v. Oracle America, Inc., where the Court held that Google’s copying of preexisting code to create new products was fair use because it facilitated programmers’ new expression. The government contends that AI training presents “an even stronger case for fair use”.

On the fourth factor — market effect — the DOJ rejected arguments that LLM training harms the potential market for copyrighted works, distinguishing between “some loss of sales” and “significant substitutive competition”. The government criticized the 2025 Kadrey v. Meta Platforms decision for improperly conflating training and outputs and treating generalized competition as market harm.

Competition, Innovation, and the Public Interest

The filing warns that an adverse fair-use ruling “would hamper competition in the market for LLMs, because only the largest technology companies might have the capital necessary to pay licensing fees”. Such fees, the government argues, “would disproportionately benefit legacy media outlets due to the sheer volume of their written publications”.

“It is not in the public’s interest for the largest technology companies to have an oligopoly on LLM training due to licensing entry barriers that function primarily as large subsidies for old mainstream media companies,” the brief states.

The DOJ also highlighted how LLMs can “level the playing field between mainstream and independent publishers” — for example, by enabling authors with limited resources to generate images to accompany articles or by directing users to dissenting sources. The government noted that even New York Times authors are using LLMs to help them “conceptualize and edit” articles.

New York Times Responds

The New York Times, which sued OpenAI in December 2023 alleging that the company “stole billions of dollars’ worth” of its journalism, sharply criticized the administration’s intervention. Graham James, a spokesperson for the newspaper, said in a statement: “The Administration is siding with a handful of trillion-dollar AI companies at the expense of the countless American creators whose work they stole”. James added: “Both AI and creators can thrive — AI companies simply need to pay fairly for the content that makes their products possible, as copyright law requires”.

The Times is seeking monetary damages, a court order preventing companies from training their models on its material, and the destruction of any models trained on its work.

Legal Context and Procedural Posture

The filing comes as dueling motions for summary judgment in the case were due to be filed on September 4. OpenAI and Microsoft moved for summary judgment in the case involving 10.8 million articles, citing only 24 verbatim outputs out of 20 million chat logs.

The DOJ’s position contrasts with the May 2025 report from the U.S. Copyright Office, which treated the application of fair use to generative AI training as a fact-specific inquiry depending on the works used, the source of training material, the purpose of the model, and its outputs. The government’s filing signals that AI training has become “both a copyright-law issue and an Administration-level policy priority”.

Conclusion

“The fair-use inquiry hinges on the specific facts and uses at issue in each case,” the government acknowledged. “But it would be problematic — and legally incorrect — to impose broad copyright liability that would generally render training of AI models impermissible without licensing”. The DOJ concluded that LLM training is “consistent with that creative ‘progress’ that is the basic constitutional objective of copyright itself”.

Attachment: Statement of Interest of the United States