OpenAI has put up a web post titled "Fair use benefits everyone", made up of excerpts from court filings that argue training language models on copyrighted works is lawful fair use. It appeared on Friday 30 October, the day amicus briefs were due in the consolidated copyright litigation against the company in the Southern District of New York, where Judge Sidney Stein presides. The plaintiffs are The New York Times and other news publishers, Ziff Davis, and a group of authors led by the Authors Guild, who are seeking class certification. Briefs for the other side came from publishers and creative-industry groups, who say fair use does not stretch to AI training.
“Copyright is not a general remedy for technological disruption or a means for reallocating the gains from a new technology.”
Public Knowledge, in its amicus brief · ipwatchdog.com ↗
The outlets disagree on how much the post gathers: IPWatchdog counts 20 filings, all but one of them amicus briefs, while Winzheng counts 12 filers. The supporters are a varied crowd. They include library and disability groups such as the American Council of the Blind, tech trade bodies, law professors, computer scientists, the venture firm Khosla Ventures, former Senator Rick Santorum, former Representative Rick Boucher and the US government. The Justice Department filed a statement of interest on 1 September saying that licensing requirements for training would favour big tech firms and hurt competitiveness and national security. It also criticised the way the 2025 Kadrey v. Meta ruling handled the fourth fair-use factor, market effect.
The arguments differ by filer. The Electronic Frontier Foundation says the rightsholders want a lawful-acquisition rule with no basis in statute or case law, and objects to penalising general-purpose AI tools. Public Knowledge wants training judged separately from how data was acquired and what a model outputs, pointing to image search and Google Books. Authors Alliance argues that "market dilution" from non-infringing outputs is not market harm, and CCIA, with NetChoice and TechNet, calls training among the most transformative uses. Emory professor Matthew Sag filed a neutral brief on "expressive substitution", noting that OpenAI's early training drew on books from shadow libraries, which bears on market effect.
The other side filed too. The Association of American Publishers says fair use is no licence for courts to rewrite the framework Congress set. Representatives Raskin, Nadler and Johnson, and the Human Consent Foundation and WME, back the rightsholders, and the congressional brief disputes the Justice Department's national-security argument. SAG-AFTRA, the Writers Guild and the American Federation of Musicians filed joint briefs urging the court to reject fair use for AI training. Winzheng reads OpenAI's post as a push for public opinion: it excerpts the briefs rather than reproducing them, and the company issued no separate statement.
About OpenAI
OpenAI has faced copyright suits over its training data since 2023. In September 2023 the Authors Guild and 17 authors, among them George R. R. Martin and John Grisham, filed a class action accusing it of using their copyrighted work. The New York Times sued OpenAI and Microsoft over training data in late December 2023.
Ziff Davis, publisher of ZDNet, PCMag and IGN, sued OpenAI in Delaware federal court on 24 April 2025. Suits by media groups including The Intercept, the Times, Tribune Media and Ziff Davis were later consolidated under District Judge Sidney Stein, the case now at the centre of the amicus filings.
11
Works credited
27
Models indexed
Models on AIPopDB
The ground is unsettled. In June 2025 Bartz v. Anthropic and Kadrey v. Meta each accepted fair use on their own facts without setting a national rule, and the Third Circuit rejected a fair use defence in Thomson Reuters v. ROSS Intelligence on 29 September 2026. Cross motions for summary judgment on fair use are pending before Stein, and the earlier AI-training fair use rulings came from two California judges rather than from New York. Reply briefs are due on 20 November 2026.
Sources · 2