The Justice Department filed in support of OpenAI's fair use defence
The statement of interest arrived before summary judgment motions in the New York Times case, and days before two more newspapers filed suit. It changes the odds in every pending case at once.
The US Justice Department has filed a statement supporting OpenAI and Microsoft's argument that training models on copyrighted text constitutes fair use, according to Wired. It arrived ahead of summary judgment motions in the New York Times' lawsuit against both companies.
A statement of interest from the government is not binding on the court. It is also not nothing: it puts the executive branch's reading of a contested statutory question in front of a judge who has to decide it, and it does so in the case that every other AI copyright case is watching.
Why the timing carries weight
Summary judgment is where fair use is most often decided, because the four-factor analysis is frequently treated as a question of law once the facts about copying are established. A government filing that lands before those motions is aimed at the moment of maximum leverage.
The Times filed in December 2023. The intervening period has produced a queue: authors, music publishers, image libraries, and now newspapers. The Seattle Times and Newsday sued OpenAI and Microsoft in the Southern District of New York on 5 September, three days after the DOJ filing, arguing that generative AI is "a snake eating its own tail" that could "destroy the very organizations" producing the content it consumes.
Every one of those cases turns on the same question the DOJ has now taken a position on.
The argument on each side
The defence case is that training is transformative: the model does not store and reproduce articles, it learns statistical structure, and the output is a different kind of thing from the input. The strongest version points out that the fourth factor — market effect — is about substitution for the original work, and a model that has read a newspaper does not substitute for that newspaper the way a photocopy does.
The publishers' case is that it does now. The substitution objection was theoretical when models simply answered from parametric memory. It is concrete when an AI product retrieves, summarises and cites current news, satisfying the reader's demand without the visit that funds the reporting. That is the mechanism the Seattle Times complaint describes.
The awkward part of the newspaper suits
Microsoft and OpenAI have funded journalism projects and fellowships at the Seattle Times — a publication now suing them. A Microsoft spokesperson told GeekWire the company was "surprised by the lawsuit" but "always happy to sit down and explore solutions to this type of dispute."
That is the licensing posture in a sentence, and it is what the litigation is really about. Both companies have signed content deals with other publishers. Litigation and licensing are running as parallel tracks, and the outcome of the fair use question sets the price on the licensing track: if training is fair use, licences are optional and cheap; if it is not, they are mandatory and expensive.
A DOJ statement supporting fair use moves that price before a single case is decided.
What it does not settle
The filing addresses training. It does not obviously reach retrieval-augmented output, where a live product fetches and summarises a specific current article — the conduct that most directly substitutes for a visit, and the conduct the newspaper plaintiffs emphasise.
Courts may well split those, finding training transformative and output substitutive. That would be the worst outcome for nobody and the most likely one, and neither the DOJ filing nor the complaints are drafted as though it is coming.
Runs the newsroom. Rename this profile in the studio to your own byline.
Related
Every weekday, the AI stories that moved money or shipped code.
No cross-posting, unsubscribe anytime. See all newsletters