Wikimedia brief in OpenAI copyright case splits AI training from chatbot output
In brief
- Wikimedia Foundation filed an amicus brief Friday in the consolidated OpenAI copyright case in New York.
- LLM training may qualify as fair use under US law, the brief argues.
- Verbatim chatbot reproduction of protected work shouldn't get the same protection, the brief says.
- Publishers, including The New York Times, allege infringement by OpenAI and Microsoft.
- Court has not ruled; publisher groups and unions filed related briefs.
A middle path between two stages
The filing landed in In Re: OpenAI, Inc. Copyright Infringement Litigation, a consolidated case in the Southern District of New York (case number 1:25-md-03143). The suits pit publishers, The New York Times among them, against OpenAI and Microsoft. Those claims are allegations, and the court hasn't decided them.
Wikimedia's brief splits the AI pipeline in two. Training comes first. Output comes second.
On training, the Foundation argues that feeding copyrighted works into an LLM may qualify as fair use. Fair use is the legal doctrine that lets people use protected material without permission in certain circumstances (commentary, research or transformation, for example). Output is a different matter. According to Crypto Briefing's report on the filing, the Foundation doesn't want verbatim regurgitation of protected work by chatbots to fall under the same protection.
Why Wikipedia cares about where the line falls
The Foundation's core concern is that both sides are pushing the court toward overly broad conclusions.
That worry cuts both ways.
If the court rules that training itself is infringement, Wikipedia's editors could lose access to useful AI-based tools, the report said. On the other hand, Wikimedia sees a ruling that favors outputs as something that could discourage people from sharing knowledge online at all. For a project built on people sharing knowledge for free across the web, neither outcome looks good, so it's asking for a narrower answer on each question.
Other voices in the case
Wikimedia isn't the only outside party weighing in. Related amicus briefs from publisher groups and unions landed around the same time, per Crypto Briefing. The outlet reported that the Association of American Publishers has filed briefs opposing broad fair-use protections for AI developers, and that the US government filed a statement of interest in September 2026 supporting fair use for AI training.
The dispute isn't new. Publishers began accusing AI developers back in 2023 of using their copyrighted articles without authorization to build models, and no ruling on those allegations has been issued yet.
Frequently asked questions
What is the Wikimedia Foundation arguing in the OpenAI copyright case?
According to Crypto Briefing, Wikimedia's amicus brief splits the AI pipeline into training and output. It argues that feeding copyrighted works into an LLM may qualify as fair use, but chatbots reproducing protected work verbatim to users shouldn't get the same protection.
Why does Wikipedia care how the court rules on AI training?
The Foundation's concern is that both sides are pushing the court toward overly broad conclusions. A ruling that training itself is infringement could cost Wikipedia editors access to useful AI-based tools. A ruling favoring outputs could discourage people from sharing knowledge online, according to the report.
What is fair use?
Fair use is the legal doctrine that allows use of protected material without permission in certain circumstances, such as commentary, research or transformation. It sits at the center of whether AI training on copyrighted works is lawful.
