The foundation behind Wikipedia warns that either sweeping restrictions or blanket protections for AI could undermine access to knowledge.
The Wikimedia Foundation has urged a balanced approach to the OpenAI copyright lawsuit brought by The New York Times. In a court filing, the organisation behind Wikipedia warned that an extreme ruling could threaten the open web.
The foundation submitted an amicus brief, offering the court its perspective on the dispute involving OpenAI and Microsoft. It supports fair use protections for some AI activities, but opposes treating those protections as unlimited.
The case concerns the companies’ use of the newspaper’s copyrighted material, including for training artificial intelligence models.
OpenAI copyright lawsuit raises competing risks
The New York Times argues that AI companies should not have unrestricted access to use its published content. OpenAI and Microsoft, however, maintain that training and deploying models using published data supports technological development and qualifies as fair use.
Wikimedia shares their concern that a narrower interpretation of fair use could restrict technological innovation. However, it also warns that blanket protection could encourage a more closed web and weaken access to knowledge.
Its concern centres on the availability of secondary sources that help sustain free knowledge projects. The foundation argues that courts must protect room for new technologies while also preserving the human-created material those technologies rely on.
Foundation draws limits around AI uses
For model training, Wikimedia supports treating the collection of copyrighted material as fair use when models differ meaningfully from their sources. Its position therefore distinguishes between creating something new and simply reproducing existing work.
The brief also addresses retrieval-augmented generation, or RAG. This technique allows language models to consult authoritative information outside their training data when generating responses.
Wikimedia supports fair use for some RAG applications. However, it does not support using the technique to reproduce existing articles without attribution.
Stephen LaPorte, the foundation’s general counsel, warned that careless application of fair use principles could harm free knowledge and open-source technology. He also stressed that Wikipedia’s collaborative work depends on accessible human knowledge.
The lawsuit is one of more than a dozen similar cases moving through the courts. A summary judgment is expected soon, and the outcome could establish an important precedent.
