A sentence aimed at the Justice Department

Three ranking House Democrats filed an amicus brief on 30 October in In re OpenAI, Inc. Copyright Infringement Litigation, the consolidated case in the Southern District of New York that includes The New York Times Co. v Microsoft and OpenAI. Their argument fits in one line, and they wrote it that way: “National security is not a fair use factor under the Copyright Act.”

The filers are Jerry Nadler, ranking member of the subcommittee on the administrative state, regulatory reform and antitrust; Jamie Raskin, ranking member of the House Judiciary Committee; and Hank Johnson, ranking member of the subcommittee on courts, intellectual property, artificial intelligence and the internet.

What they are answering

The target is the Department of Justice, which intervened on 1 September to support the AI companies, arguing that restricting their use of copyrighted text would threaten American AI development and, through it, national security.

The lawmakers’ position is that the Copyright Act contains no such exception and that reading one in would add “a national security exception that Congress did not enact”. The claim is not that AI training can never be fair use. It is narrower and harder to answer: the four statutory fair use factors are the four factors, and a national interest in a technology is not among them. Section 107 lists purpose and character of the use, nature of the work, amount used, and effect on the market.

A marble-tiled hall with a large arched window, two people silhouetted as they walk away.
Twelve suits against OpenAI and Microsoft are consolidated into the proceeding. Stock image. Nemika F · pexels · Pexels License

Members of Congress filing against an executive-branch brief in a copyright case is unusual. It is also a specific kind of argument from a specific place: three legislators telling a court what the legislature did and did not write.

Wikipedia filed the same day, and sided with neither

The Wikimedia Foundation lodged its own amicus brief on 30 October, and took a position that cuts across the parties. On training, it argues that using copyrighted content to build a large language model should count as fair use when the resulting model is new and different from its sources — the defendants’ position.

On retrieval-augmented generation and similar grounding techniques, it argues the opposite: fair use should not apply, and companies offering those features should respect each publisher’s licence. For Wikipedia that licence is CC BY-SA 4.0, which requires attribution and keeps derivatives under the same terms. The Foundation also asks that scraping of its projects stay within public API rate limits or move to its Enterprise API.

Stephen LaPorte, the Foundation’s general counsel, said “the court needs to understand that Wikipedia is created through a collaborative process”. Chief executive Bernadette Meehan said “expanding access to knowledge through AI is a powerful goal”. The brief warns in both directions: an over-restrictive ruling could damage the open-source tools Wikipedia depends on, and a blanket fair use ruling could produce a more closed web with fewer secondary sources left to cite.

A wood-panelled legislative chamber with rows of desks, microphones and green leather chairs.
The lawmakers argue the four statutory fair use factors are the only four. Stock image. Héctor Berganza · pexels · Pexels License

What to watch

Whether Judge Sidney Stein takes the national security argument seriously enough to address it. An amicus brief does not have to be accepted, and the court has not said whether it will take either of these. The consolidated case covers twelve suits against OpenAI and Microsoft, and no appellate court has yet ruled on whether training on copyrighted work is fair use. The question of whether a national interest can enter that analysis at all is the one the lawmakers are trying to close before it opens.