The Wikimedia Foundation, the nonprofit behind Wikipedia, has filed an amicus brief in the consolidated copyright litigation against OpenAI and Microsoft, asking a federal court in Manhattan to craft a fair-use analysis that protects the open web — whichever way the underlying dispute is decided. The Foundation's general counsel, Stephen LaPorte, laid out the organization's position in a post published on October 30, and The Verge reported the filing under a stark framing: a bad AI copyright ruling, Wikipedia's operator argues, could threaten the open internet.
The brief was filed in the Southern District of New York in the case known as In re: OpenAI, Inc. Copyright Infringement Litigation — a multidistrict proceeding that consolidates twelve separate lawsuits, the most famous of them The New York Times Co. v. Microsoft and OpenAI. The Foundation is notably not taking sides between the newspaper and the AI company; its ask is narrower and, in its view, more fundamental. For readers tracking how the courts are reshaping AI's relationship with published content, our AI policy coverage follows every development.
Why Wikipedia Weighed In
The Foundation's stake is straightforward. Wikipedia comprises more than 67 million articles across 300 languages, built and maintained by nearly 250,000 volunteer editors. That work, LaPorte wrote, depends on access to trustworthy sources — including journalism and published research — that volunteers can read, verify, and cite.
AI, meanwhile, leans on Wikipedia in return. The encyclopedia is one of the largest and highest-quality datasets used to train large language models, and it also powers search engines and voice assistants. "A copyright ruling that disrupts either side of that relationship," the Foundation argues, "could affect how everyone finds and uses information."
The Question Before the Court
The New York Times sued OpenAI and Microsoft in 2023, alleging the companies used its copyrighted journalism without permission to train and operate their AI systems. OpenAI and Microsoft contend that using publicly available material to develop AI systems qualifies as fair use — the legal doctrine that permits certain uses of copyrighted works without a license.
The Wikimedia Foundation's brief warns against answering that question too broadly in either direction.
If courts interpret fair use too restrictively, the Foundation argues, AI development would become the province of the largest corporations that can afford content licensing at scale — while open-source developers, unable to pay, would be locked out of the field entirely. That outcome, it says, would directly harm Wikimedia volunteers' ability to access knowledge and experiment with the open-source tools they use to build and maintain the encyclopedia.
But a ruling that hands AI companies a blanket exemption from copyright rules carries its own danger. If developers can ingest and republish copyrighted material without attribution or licensing, publishers have a powerful incentive to lock their work behind technical and legal barriers. The result would be fewer open sources available to the volunteers who write Wikipedia — and reduced visibility for the very publishing ecosystem, from newsrooms to researchers, that both Wikipedia and the public depend on.
Training Versus Answering
The most technically interesting part of the Foundation's argument is its insistence that not all AI uses of copyrighted works are the same. The brief draws a line between training a model — where the system learns patterns from large collections of material, and where the Foundation says fair use may properly apply when the resulting model is genuinely new and different from the works it learned from — and using retrieved source material to answer a user's question, a use it says deserves separate analysis.
Some AI products, the Foundation notes, answer questions by surfacing retrieved text in ways that can substitute for the original work. The Foundation's position is that courts should evaluate how material is actually being used, rather than treating every interaction between AI and copyright as one identical question.
Stakes Beyond One Courtroom
The Wikimedia Foundation is far from the only interested party filing in the multidistrict litigation, but its brief is distinctive: it asks the court to protect the knowledge commons from both litigants' most expansive arguments. The outcome will shape not just damages in a dozen lawsuits, but whether the next generation of open-source AI models can be legally trained by anyone other than the wealthiest companies — and whether the open sources that feed human knowledge online stay open at all.
For the volunteers who maintain Wikipedia, and for every organization that relies on a searchable, citable web, the foundation of the argument is simple: fair use should keep working — for training, for reference, and for the public. The filing arrives at a congested moment for AI copyright law, as courts begin testing how doctrines written for search engines and photocopiers apply to systems that learn from what they read.
---
Get the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
