Amicus Briefs Flood the OpenAI Copyright Case as "Market Dilution" Becomes the Key Fight
Photo: Artem_Apukhtin / Pixabay
The consolidated copyright litigation against OpenAI and Microsoft in New York has drawn a wave of friend-of-the-court briefs this week. The briefs arrive as Judge Sidney Stein of the Southern District of New York weighs cross-motions for summary judgment that could settle much of the dispute. Amicus filings were due on 30 October 2026, a week after the parties filed their oppositions on 23 October.
The multidistrict litigation, In re OpenAI, Inc. Copyright Infringement Litigation, brings together suits from writers (the Authors Guild among them, plus a putative class) and from publishers including The New York Times, the Daily News and Ziff Davis. The plaintiffs say that copying their works to train OpenAI's models was infringement. OpenAI says the training was fair use.
Who weighed in
Law360 reported a wave of proposed amicus briefs from both sides, filed by publishing, news and creative-industry groups, public-interest and technology organisations, and legal academics. Authors Alliance, which filed one of them, said organisations and individuals on both sides had already filed briefs and that more were expected in the following days. According to Law360, a common thread is a request that the court treat three things as separate acts rather than folding them into a single fair-use ruling: training the models, acquiring the copyrighted material, and the answers the models produce.
The "market dilution" question
Authors Alliance filed for leave on 29 October. Its brief supports the defendants but addresses only one issue: whether "market dilution" counts as a harm that copyright recognises. The theory holds that AI training damages the market for the original works because a model can then turn out large numbers of new, non-infringing works that compete with them. It traces back to Judge Vince Chhabria's 2025 ruling in Kadrey v. Meta. In Bartz v. Anthropic, Judge William Alsup turned down a comparable argument, while the US Copyright Office aired a form of the idea in a draft report on AI training released ahead of formal publication. The author plaintiffs in New York have asked Judge Stein to adopt it.
Authors Alliance argues that the fourth fair-use factor reaches uses that replace an author's protected expression, not fresh works that simply chase the same audience. On its account, scale cannot turn ordinary competition into a copyright injury, and the theory would in effect protect ideas, facts and genres that the Copyright Act leaves free. It also says the theory would fall hardest on scholars, critics and biographers who build on earlier works. The group took no position on how OpenAI obtained its data or on whether particular outputs infringe.
The Electronic Frontier Foundation filed its own brief the same day. It focuses on two plaintiff arguments: that potential competition from AI output makes training unfair even if the use is transformative, and that scraping or downloading without permission, or from sites hosting pirated copies, is infringing on its own or taints later uses. EFF argues that a lawful-acquisition requirement would cast doubt on the routine copying behind everyday web browsing and on the scraping that researchers and journalists rely on. It also argues that mandatory licensing would favour the largest companies.
What to watch
The briefs show where the fight now sits. Earlier rulings on AI training have tended to accept that training is a transformative purpose, so the battle has moved to the fourth factor and to how the training data was obtained. If Judge Stein accepts market dilution, plaintiffs would gain a route to defeat fair use even where no output copies their work. If he rejects it and treats acquisition as a separate question, the case may turn on piracy-sourced datasets, the issue Judge Alsup left for trial in Bartz before Anthropic agreed to settle that case in 2025.
Any ruling will bind only this case, and an appeal to the Second Circuit is likely either way. Still, a decision in the largest AI copyright proceeding in the US will be read closely by courts and policymakers elsewhere.
An Indonesian angle
Indonesia's Copyright Law (Law No. 28 of 2014) has no open-ended fair-use clause. It lists specific uses that do not infringe, provided the source is cited and the use does not unreasonably prejudice the rights holder. No Indonesian court has yet ruled on AI training. The US debate over whether training, data acquisition and outputs should be judged separately is still a useful framework for Indonesian developers and rights holders, especially as the government discusses rules on AI and content licensing. Companies training models on Indonesian content should keep records of their data sources now, because the question of how the data was obtained is the one gaining ground in the US.
Sources
Do you or your clients need trademark, patent, annuity or industrial design work in Indonesia? We handle filings, prosecution and renewals, with updates in English. Email us for a quote.
Contact UsIf you enjoyed this article on our portal, you can buy me a coffee via the PayPal link below.
PayPal
