The Wikimedia Foundation filed an amicus brief on Friday to intervene in the ongoing copyright lawsuits between publishers like The New York Times and technology firms OpenAI and Microsoft. The organisation argues that a binary legal decision could produce two damaging outcomes. If courts rule that large language models cannot train on copyrighted material, Wikipedia loses access to essential AI tools that help editors verify facts and summarise complex topics. Conversely, if the law permits models to train on such work but also allows them to reproduce that material directly in user responses, the open web faces threats of widespread text duplication.
This intervention highlights a specific tension between training data rights and output restrictions. The brief distinguishes between the internal use of copyrighted text to improve model performance and the external generation of content that mimics original works. By advocating for this distinction, the foundation seeks to prevent a ruling that either cripples its own research capabilities or enables automated plagiarism on a massive scale.
- The brief was filed on Friday in the US federal court system.
- OpenAI and Microsoft are defendants in the publisher-led lawsuits.
- The Wikimedia Foundation is taking a neutral stance rather than supporting either side.
