What the Brief Says
The Trump administration filed a 20-page brief in Manhattan federal court on 2 September 2026 arguing that training large language models on copyrighted material generally qualifies as fair use. The filing supports OpenAI in its dispute with The New York Times and other newspapers over the unlicensed use of their reporting to train the models behind ChatGPT. It is the administration's first intervention in any of the pending AI training cases.
The brief states that the United States has "a strong interest in continuing to develop a robust and competitive artificial intelligence industry that sets the standard for the practice and procedure of AI use globally." Fair use, in that framing, functions as a competitiveness instrument: an administration that wants American labs training at scale has reason to oppose licensing requirements as a precondition. The argument subordinates copyright doctrine to industrial policy, a positioning rights holders across the broader litigation will contest.
The word "generally" carries weight. The administration did not argue that every training use qualifies as fair use, only that the practice defaults to it, leaving courts room to find specific uses excessive. TechCrunch reported the brief as a defense of OpenAI's unlicensed use specifically, which suggests the government places the ChatGPT training runs inside the default rather than at its edge. Whether the court reads that distinction the same way is the operative question.
First Federal Intervention in the AI Copyright Wave
RTÉ reports the filing as the first time the United States government has weighed in on the wave of copyright cases brought over AI training. The plaintiffs in that wave include authors, publishers, music labels and news outlets. Until this filing, the executive branch had left the fair use question to district courts working through the doctrine case by case, with no stated federal position on the central legal issue.
A government brief functions as a signal to other courts. Judges deciding the remaining cases read the same filings, and a stated executive position on fair use gives defendants a citation carrying institutional weight beyond its legal force. That suggests the administration intends its position to travel beyond one Manhattan courtroom, whether or not it files again. The open question is how much weight trial judges give executive-branch views on a doctrine courts have historically owned.
The available reporting does not reproduce the brief's full legal argument, so claims about its doctrinal detail — whether it addresses memorization, output regurgitation, or market harm analysis — cannot be verified here. What the record establishes is the position's breadth: fair use as the general rule for training, stated in a 20-page filing by the federal government on the side of a defendant rather than the plaintiffs.
What It Means for the NYT Lawsuit
For The New York Times, the brief changes the political context of its case without changing its legal posture. The newspaper's claims — that OpenAI used its work without permission to train commercial models — remain before the same Manhattan federal court. The government's position now sits on the defendant's side of the docket, filed by an administration that frames AI industry growth as a national priority.
Briefs of this kind do not decide cases. They supply arguments a court may adopt and a record of executive-branch thinking that appellate panels can consult. The practical effect is asymmetrical: OpenAI gains a government-endorsed fair use framework to argue from, while the Times must litigate against both a well-resourced defendant and the stated policy of the federal government. The case's trajectory will be set by its own record, not by this filing.
Two signals matter from here. First, whether the court's eventual ruling adopts the brief's general fair use framing or narrows it to the specific facts of news training, where market harm arguments run strongest. Second, whether other copyright plaintiffs — the authors, publishers and music labels in parallel suits — amend their theories to target uses the government's stated position does not reach. Neither outcome is determined by the filing itself.
Sources
- techcrunch.com — US Government Sides with OpenAI on Training LLMs on Copyrighted Material
- sjvsun.com — Government sides with OpenAI in New York Times ... (2026-09-02)
- realitytea.com — Donald Trump Administration Sides With OpenAI Against Media Giants (2026-09-02)
- rte.ie — US government backs OpenAI in NY Times copyright case (2026-09-02)
- New Horizon daily digest, 2026-09-03 (topic provenance)