US Justice Department files first-ever statement of interest backing AI-training fair use in NYT v OpenAI

What happened

On September 1, 2026, the U.S. Department of Justice filed a 20-page statement of interest under 28 U.S.C. § 517 in *In re OpenAI, Inc. Copyright Infringement Litigation* (MDL No. 25-md-3143), the consolidated federal cases before Judge Sidney H. Stein in the Southern District of New York that include The New York Times' copyright suit against OpenAI and Microsoft. It is the first time the federal government has formally intervened in any AI-training copyright case.

The brief argues that training large language models on copyrighted text is "transformative—spectacularly so," because models convert works into statistical relationships rather than reproducing their expressive content. It draws a three-stage line through the AI pipeline — collecting copyrighted works, training a model, and generating outputs — and defends only the copying that happens during training; claims about how training material was acquired and about outputs that reconstruct protected expression are left as open questions.

The DOJ further argues that requiring licenses for training data would create entry barriers that "favor large established media companies," let a handful of publishers set prices only the biggest AI developers could afford, concentrate the LLM market, and disadvantage U.S. companies against foreign rivals — tying the position to national security and economic competitiveness. If publishers want to be paid, the brief says, Congress should create a licensing framework rather than courts creating one by accident.

Why it matters to creators

The government is now formally on record saying AI training on copyrighted text should be fair use — the central legal argument creators and publishers have been fighting in dozens of lawsuits. If Judge Stein adopts this reasoning, the training-licensing argument that authors, news organizations, and music labels have been advancing loses its most viable legal pathway. The brief also implicitly argues against a general creator-licensing market for training data, on the ground that it would entrench incumbents on both sides.

What is not yet known

A statement of interest is non-binding; the court is not required to defer to it. The New York Times condemned the intervention as prioritizing "trillion-dollar AI companies" over creator compensation. The DOJ did not say whether it will file parallel statements in the music-industry cases against Anthropic, Suno, and Udio. Above the Law also reported an undisclosed context: the Trump administration has reportedly been in discussions about accepting an approximately $42.6 billion equity stake in OpenAI, one of the defendants the brief supports.