The US Department of Justice has entered the most consequential copyright fight of the AI era, urging a federal judge to rule that training artificial intelligence models on copyrighted books, articles and photographs is fair use — and therefore lawful without a license.
In a statement of interest filed in the US District Court for the Southern District of New York, the department argued that The New York Times' position in its lawsuit against OpenAI and Microsoft is "inconsistent with basic copyright law principles" and, if adopted by the court, could "severely hamper 'the Progress of Science and useful Arts.'" The filing, reported by Reuters, The Wall Street Journal and The Washington Post, puts the weight of the US government behind the argument that training AI models "in and of itself" does not violate copyright law. It is the clearest signal yet of where Washington stands as the AI industry's legal battles reach a decisive stage.
What the Justice Department Told the Court
The Times sued OpenAI and its largest investor, Microsoft, in 2023, accusing them of "systematic and competitive infringement" for using millions of its articles to develop commercial AI models without permission or payment. After other publishers filed related lawsuits, the cases were consolidated last year in the Southern District of New York.
In its filing, the Justice Department points to Section 107 of the Copyright Act of 1976, which allows unlicensed use of protected works for purposes such as criticism, comment, news reporting, teaching, scholarship and research, provided the use is weighed against statutory factors. The government's core contention is that any use by the AI companies should be considered fair use for a transformative purpose.
To make that case, the department leans on a 2015 Second Circuit ruling in Authors Guild v. Google, which held that "the more the appropriator is using the copied material for new, transformative purposes, the more it serves copyright's goal of enriching public knowledge" and the less likely the use substitutes for the original work.
The Times Fires Back
The newspaper has rejected the transformative-use argument outright. According to its filings, there is "nothing transformative about using its content without payment to create products that substitute for The Times and steal audiences away from it." The publisher argues that OpenAI's generative models "compete with and closely mimic the inputs used to train them," which it says nullifies any fair use defense.
A spokesman for the Times, Graham James, said the government was "siding with a handful of 'trillion-dollar A.I. companies' at the expense of American creators," as The Washington Post reported. The irony has not gone unnoticed that the Times itself uses generative AI in its newsroom — for digitally voiced articles, translations and, as the paper told staff, to "help us report stories we couldn't otherwise," according to Semafor.
Legal Experts Are Split
Public-interest lawyers and academic commentators disagree sharply on who has the better of the argument.
Tori Noble, a staff attorney at the Electronic Frontier Foundation, told Reason that training AI models on copyrighted works is "squarely what transformative use is." Large language models, in her view, are "general purpose tools" and "exactly the kind of creation that copyright is designed to foster."
Satya Marar, a research fellow at the Mercatus Center at George Mason University, offers a more measured reading: there is "no blanket fair use exception for training generative AI," he said, but the training itself is "highly likely to be deemed transformative" by the court.
On the other side of the ledger, Nathan Goodman, also of the Mercatus Center, notes that a ruling in the Times' favor would "substantially raise the direct financial costs and the transaction costs of forming a training data set," which he believes would "impede iterative innovation that builds on prior existing knowledge." Noble adds a caution of her own: there is a "real risk in these cases that courts will misunderstand the technology" and issue rulings grounded in perception rather than in how the systems actually work.
What Happens Next
The statement of interest does not bind the court, but it signals that the US government will argue forcefully for a broad reading of fair use as the case moves toward judgment. The Supreme Court's 2023 decision in Andy Warhol Foundation v. Goldsmith — which framed copyright's purpose as promoting progress "without diminishing the incentive to create" — looms over the analysis either way.
With several copyright lawsuits against AI companies still pending, this week's filing is unlikely to be the last time the government weighs in. Whatever the Southern District of New York decides, the ruling is expected to reshape how — and at what price — AI companies assemble training data for years to come.
Stay Ahead of AI
The legal fight over AI is moving fast. Get the latest AI developments in one place.
Read more AI news →
