The U.S. Government Sides With OpenAI in the Copyright Fight

For the first time, the U.S. government has formally stepped into one of the major copyright lawsuits against AI companies – on OpenAI’s side. The Justice Department told a federal court in New York that training large language models on protected text is covered by the U.S. fair use doctrine and serves the national interest. The New York Times, which started the case nearly two years ago, calls it a move to favor trillion-dollar companies at the expense of creators.

Key takeaways

• On September 2, 2026, the U.S. Justice Department filed a so-called statement of interest with the federal court for the Southern District of New York, without becoming a party to the case.
• Its core claim: AI training transforms other people’s text enough to create new, permitted material, and the public benefit outweighs any competitive harm.
• The government invokes national security, arguing that a U.S. lag in AI would benefit “foreign adversaries.”
• It is the first time the U.S. government has taken a formal position in any of the many AI copyright cases.
• Europe operates under different rules: no fair use, but a statutory exception for text and data mining with an opt-out right for rights holders.

What the Justice Department filed

The dispute is before the federal court for the Southern District of New York, where Judge Sidney H. Stein has consolidated more than a dozen lawsuits by publishers and authors against OpenAI and Microsoft into one proceeding. The New York Times filed its complaint on December 27, 2023, accusing OpenAI of using millions of its articles without permission to train ChatGPT. It is seeking damages in the billions; under U.S. law, willful infringement can carry up to $150,000 per work.

A statement of interest is a brief that a government agency uses to tell the court its legal view without being a plaintiff or defendant. Associate Attorney General Stanley Woodward Jr. called the step a “historic” filing and pointed to President Trump’s line that he views “AI dominance as critical to promote national security, prosperity, and economic mobility for all Americans.” The department asks the judge to rule for OpenAI in the pending motions for judgment without a trial – the deadlines for which fall in these days.

The arguments: fair use and national security

On the merits, the government relies on the “transformative use” doctrine in U.S. copyright law: if a use creates something fundamentally new, it can be permitted as fair use even without consent. Training “sufficiently” transforms the text into new material, the brief says. The “creative possibilities and public benefits” of large language models “far outweigh any competitive harm.” A ruling for the New York Times would “thwart creative and scientific progress” and harm American prosperity.

The link to security policy is new. AI models could help security agencies “draw inferences from real-world facts such as those conveyed in New York Times articles” and matter for intelligence analysis, weapons systems, and targeting recommendations in combat. A U.S. AI sector slowed by a court ruling would, by this logic, benefit “foreign adversaries.” There is also an economic argument: only large companies could afford licensing fees, while free AI tools give independent authors a “level playing field” against large media houses. The New York Times counters that the government is siding with “a handful of trillion-dollar AI companies at the expense of the countless American creators whose work they stole,” and is undermining the basis for human-created content. OpenAI faces added pressure because the court already ordered it in January 2026 to hand over 20 million anonymized ChatGPT logs for discovery; earlier motions by the publishers sought sanctions over deleted data. The case is part of a widening front, as the lawsuits by Sony and Warner against Anthropic also show.

Why this counts only partly for Germany

A fair use win in the U.S. could not be transferred to Europe without more, because there is no fair use here. Germany and the EU work with fixed statutory exceptions. For AI training, the key one is Section 44b of the Copyright Act, the text and data mining exception from the 2019 EU Copyright Directive. It permits automated analysis of lawfully accessible works – but only as long as the rights holder has not reserved that right. For content available online, this objection must be machine-readable. The first German ruling on this involved photographer Robert Kneschke, whose suit against the dataset association LAION was dismissed by the Regional Court of Hamburg in September 2024; the Higher Regional Court of Hamburg confirmed the decision in late 2025, relying in part on the research exception in Section 60d. On top of that, the EU AI Act has required providers of large AI models since August 2025 to maintain a copyright policy, honor objections, and publish a summary of their training data. The European debate is therefore less about whether than about how effectively an opt-out must be implemented technically.

What happens next

The Justice Department’s filing does not bind the judge; he can weigh it or set it aside. In the short term, the decisive step is the motions for judgment without a trial, on which Judge Stein is likely to rule in the coming months; a possible jury trial would not be expected before 2027. What is already clear is that the training-data question has become a political one: in the U.S., the government defends free training as a competitive advantage, while European courts and the AI Act aim for a balance through opt-out rights and transparency. For creators in Germany, that mainly means recording their rights reservations cleanly in technical terms – a U.S. ruling, however it turns out, will not take that task off their hands.