The Pulse
DOJ Sides With OpenAI in New York Times Copyright Fight
The Justice Department is backing OpenAI and Microsoft in their copyright fight with The New York Times, according to Axios' reporting on a government filing and sources familiar with the matter.
AI.info Team ·
Axios reports that the Justice Department is backing OpenAI and Microsoft in their copyright fight with The New York Times, arguing that the use of copyrighted works to train large language models can qualify as fair use. The filing surprised the U.S. Copyright Office and the U.S. Patent and Trademark Office, two agencies closely tied to copyright policy, according to sources who spoke with Axios.
The department filed a statement of interest in the Times’ lawsuit, which accuses OpenAI and Microsoft of copying millions of articles to train systems including ChatGPT and Copilot. The filing does not make the government a party to the case and does not bind U.S. District Judge Sidney Stein, who is handling the litigation in the Southern District of New York. It does, however, put the federal government behind the companies’ central legal argument at a consequential stage of the case.
Washington backs transformation, not every output
The Justice Department’s position is narrower than a blanket declaration that every use of copyrighted material by an AI system is lawful. Axios reports that the government argues training itself should be treated as fair use because the process is new and transformative, while warning that model outputs are not automatically covered by the same reasoning.
That distinction matters to the Times, whose lawsuit attacks both the copying involved in model development and the possibility that generated answers can substitute for the newspaper’s reporting. The Times says OpenAI and Microsoft used its work without permission and that AI-generated responses can reproduce or displace the value of original journalism.
In a September 2 report, The Washington Post reported that Associate Attorney General Stanley Woodward wrote on social media:
“This Administration will never let our Nation be at a disadvantage relative to our foreign adversaries based on a plainly incorrect understanding of copyright law.”
Woodward is one of the attorneys who signed the filing. The government’s position frames AI development as tied to national security and argues that restricting training could impede scientific and creative progress.
The licensing fight sits beneath the legal argument
The Justice Department also rejects the idea that AI companies should generally need licenses before training on copyrighted material. The Washington Post reported that the government argued licensing fees could “hamper competition in the market for LLMs” by making it harder for smaller companies to build models. The filing said an industry dominated by companies able to afford broad licensing agreements would create an oligopoly.
Publishers dispute that account. Graham James, a spokesperson for The New York Times, said:
“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 that AI companies and creators could both succeed if developers paid fairly for the content used to build their products.
Axios also reports that the Justice Department’s brief says enforcing a licensing system would be cumbersome without taking a position on whether such a system could work. That combination has drawn criticism from publishers, including the argument that the government is treating copyright enforcement as an obstacle while leaving the underlying compensation question unresolved.
Agencies decline to explain the split
The Copyright Office and the Patent and Trademark Office declined to comment on the filing, while the Justice Department did not respond to Axios’ request for comment. Axios reports that no career antitrust attorneys signed the statement alongside the department’s senior officials, an unusual feature for a filing intended to present the government’s position in a private case.
The lack of coordination is politically significant because the copyright question reaches beyond OpenAI. News organizations, authors, database companies and other rights holders are pursuing claims against AI developers over the use of books, articles, code and online material in training datasets. Courts must decide whether the act of converting protected expression into model parameters is sufficiently different from the original use, and whether the resulting products compete with the works that supplied the data.
Recent rulings in separate cases involving Anthropic and Meta have strengthened AI companies’ fair-use arguments in some circumstances, particularly when plaintiffs could not show that model outputs reproduced protected expression. Those decisions do not settle the Times case, which involves different evidence, claims about market substitution and the Second Circuit’s copyright precedent.
Judge Stein still has the final word
The Justice Department’s statement is persuasive advocacy, not a ruling. Judge Stein must still assess the record assembled by the Times, OpenAI and Microsoft, including what material the companies used, how their systems reproduce text and whether the models compete with the newspaper’s original market.
Axios characterizes the filing as part of a broader pattern in which the Trump administration has intervened in media-related legal disputes. For OpenAI and Microsoft, the immediate benefit is clear: the government has supplied a national-security and competition rationale for rejecting a broad licensing requirement. For the Times and other publishers, the filing raises the cost of persuading the court that uncompensated training is a form of market substitution rather than a transformative use.
The next decision belongs to the federal court in New York, not the Justice Department. Until Judge Stein rules, the government’s position changes the political pressure around the case but does not decide whether OpenAI’s training practices violated copyright law.