Wikimedia Files AI Copyright Brief in OpenAI Lawsuit
Wikimedia filed an amicus brief in the New York federal copyright cases against OpenAI and Microsoft, urging the court to distinguish model training from chatbots that retrieve and reproduce licensed material.
By RobotPulse Newsroom ·
Watch the RobotPulse Short on YouTube
The Wikimedia Foundation has filed a friend-of-the-court brief in a major New York copyright case involving OpenAI and Microsoft. It urges the court to distinguish between training a model on copyrighted material and using a chatbot to retrieve and reproduce source text in an answer.
What happened
Wikimedia says it filed the brief on October 30 in the Southern District of New York case *In re: OpenAI, Inc. Copyright Infringement Litigation*. The proceeding combines 12 lawsuits against OpenAI, including the widely known case brought by *The New York Times* against Microsoft and OpenAI. The Times alleges that the companies used its copyrighted material without permission, including to train AI systems. OpenAI and Microsoft argue that using publicly available material to develop and operate their models can qualify as fair use (Wikimedia Foundation; The Verge).
An amicus brief lets an organization that is not a party to a case offer arguments for the court to consider. Wikimedia says it is not taking the side of OpenAI or The New York Times. Instead, its filing asks the court to consider how copyrighted material is used at different stages of an AI system.
For model training, Wikimedia argues that using copyrighted works may be fair when the resulting model is something new and different from the original sources. For retrieval-augmented generation, or RAG, the foundation draws a different line: when a chatbot retrieves material from the web and returns that material to a user, companies should follow the source’s licensing requirements. Wikipedia content is distributed under the Creative Commons CC BY-SA 4.0 license, which requires attribution and requires adaptations to remain under the same license (Wikimedia Diff).
Why it matters
Wikipedia is both a widely used source of information for AI systems and a website whose editors and infrastructure depend on people continuing to visit and contribute. Wikimedia says a ruling that is too restrictive could make it harder for open-source developers to use public information, while a blanket fair-use rule could weaken sources and communities that make knowledge available online.
The brief highlights a distinction that is becoming important in copyright disputes: learning from a work during training is not the same as retrieving and presenting that work as part of a chatbot response. Wikimedia argues that the second use can substitute for the original source and should respect licensing and attribution requirements. Its position is an argument to the court, not a ruling that establishes the law.
The underlying case could influence how U.S. courts assess copyrighted works used to train and operate large language models. The filing also makes the debate relevant beyond publishers and AI companies, because it addresses the reuse of openly licensed knowledge and the rules that apply when AI products surface it directly to users.
What to watch next
The court has not decided the fair-use question in this case. The next significant development will be how the judge weighs the parties’ arguments and whether the eventual ruling treats model training and retrieval-based answers differently.
Wikimedia says AI companies using Wikipedia material through RAG or similar grounding methods should provide attribution under CC BY-SA 4.0. It also says companies collecting or retrieving its content should follow public API rate limits or use Wikimedia’s Enterprise API. Those are positions advanced by the foundation; the court’s eventual treatment of them remains to be seen.