Wikimedia Foundation Files Brief in AI Copyright Lawsuit, Urging Balanced Approach

The Wikimedia Foundation has filed an amicus brief in ongoing copyright litigation between publishers like The New York Times and AI developers including OpenAI and Microsoft. The organization argues for a nuanced legal approach, suggesting that training large language models on copyrighted works should qualify as fair use, while chatbots directly regurgitating that protected content should not.

According to the legal filing, the nonprofit organization is concerned that absolute rulings could either hinder open-source technologies or harm the open web. Wikimedia relies heavily on human-generated content and serves as a major data source for machine learning training, yet it also highlights that AI tools can divert web traffic away from the communities maintaining these sites.

Key Arguments and Proposed Compromise

• Training vs. Output: The foundation suggests that using copyrighted material for pre-training AI models should be considered fair use when it produces something transformative.
• Licensing Requirements: If an AI service reproduces material directly to answer user queries and supplants engagement with original sources, it should respect licensing terms and provide proper attribution.
• Internal AI Benefits: Wikimedia notes that AI models also assist its editors with tasks such as fighting vandalism and detecting policy violations.

Practical Implications and Verification

Legal outcomes in these ongoing cases could shape how AI developers source training data and how publishers protect their intellectual property. Observers can review court dockets and read the official Wikimedia Foundation blog post and amicus brief to verify the organization's stated positions.