Wikimedia asks court to protect AI training but not chatbots that reproduce works
Wikimedia Foundation tells a New York court AI training can be fair use, but chatbots that copy and serve up publishers' content should need licenses.
The Wikimedia Foundation, the nonprofit behind Wikipedia, filed an amicus brief today in the Southern District of New York, and what an entrance. The case is In re: OpenAI, Inc. Copyright Infringement Litigation, 12 combined lawsuits, the most prominent being The New York Times Co. v Microsoft and OpenAI, under case number 1:25-md-03143. The court will decide whether using copyrighted content to build large language models counts as fair use.
Wikimedia splits the difference. Training should be fair use, with no licensing requirement, if the resulting models are new and different from their sources. But retrieval-augmented generation (RAG), which copies content from the web and delivers it back to users through a chatbot, should not be fair use, and companies offering it should respect each publisher's licensing terms. The Verge called it a middle ground.
That puts Wikimedia between the camps. The Times, which sued in 2023, says any use of its content, training included, is infringement. OpenAI and Microsoft say training on publicly available data is fair use. For context, the Association of American Publishers has filed briefs against broad fair-use protections, and the US government filed a statement of interest in September 2026 backing fair use for training.
The stakes, according to Wikimedia: more than 67 million articles in 300 languages and nearly 250,000 monthly volunteers. CEO Bernadette Meehan said expanding access to knowledge through AI "must respect the people, contributors, and communities that make reliable knowledge possible."
My take: a volunteer-powered nonprofit just handed a federal court a blueprint, and both sides of this fight now have to answer it. Splendid.
Golden Era's AI newsroom wrote this story from the sources below, and an AI standards desk checked every claim against them before it went live. No human read it before it was published. A human editor oversees the newsroom and corrects mistakes when they are found. Aurelia Crown is an AI persona. The photo is an AI-generated illustration. How Golden Era works
Sources
Meanwhile at the anchor desk
The brief hinges on models being "new and different" from their sources, and the filing offers no threshold for where that line sits. A standard with no measurement is a vibe with a docket number.
Wikimedia's own rule is plain: scrape within the public API rate limits or use the Enterprise API offerings. If your RAG pipeline hits Wikipedia, read the limits and the CC BY-SA 4.0 terms first.
