# Wikimedia Foundation Files Amicus Brief in OpenAI Copyright Litigation The foundation argues for a nuanced fair-use distinction between model pre-training and retrieval-augmented generation. - **Source:** [Unite.AI](https://www.unite.ai) - **Published:** 2026-10-30 - **Topic:** [Business](https://brifing.si/c/business) - **Original article:** - **On Brifing:** https://brifing.si/s/51ed9e38-4a70-41db-b404-49187c918970 ## Legal Intervention Strategy On October 30, 2026, the Wikimedia Foundation requested leave to file an amicus brief in the consolidated copyright litigation against OpenAI and Microsoft. The motion, submitted to Judge Sidney H. Stein, aims to influence the court’s interpretation of fair use regarding large language model development and the use of copyrighted data. ## Distinguishing Training from Retrieval The Foundation proposes a bifurcated legal approach, arguing that model pre-training constitutes fair use because it is transformative and non-substitutive. Conversely, it contends that retrieval-augmented generation (RAG) fails the four-factor fair-use test because it directly competes with original sources by copying and presenting content chunks to users for informational purposes. ## Impact on Open Source Wikimedia warns that overly restrictive fair-use rulings could consolidate AI power within the largest corporations, effectively locking out open-source developers. However, it also cautions that blanket fair use might incentivize publishers to erect paywalls, ultimately weakening the open web and reducing the visibility that Wikipedia relies on for volunteer recruitment. ## Technical and Licensing Stance The brief distinguishes between technical scraping and copyright use, suggesting that automated data acquisition should be governed by terms of service rather than copyright law. It emphasizes that RAG systems should respect existing licensing frameworks, such as the Creative Commons CC BY-SA 4.0 license, which mandates attribution and share-alike requirements. ## Precedent and Analytical Framework Citing the 2023 Supreme Court decision in Andy Warhol Foundation v. Goldsmith, the Foundation argues that each distinct stage of AI development requires a separate fair-use analysis. It also references the 2025 Bartz v. Anthropic ruling to support the necessity of evaluating pre-training and RAG as separate, independent technical processes. ## Operational Strain on Platforms Non-human traffic from aggressive scraping currently strains Wikimedia’s infrastructure, forcing the organization to implement technical safeguards that sometimes hinder human access. The Foundation maintains that AI developers must respect public API rate limits or utilize the Wikimedia Enterprise API to ensure the sustainability of its thirteen free-knowledge platforms. ## Internal AI Utilization The Foundation highlights that it already employs machine-learning tools to combat vandalism, detect copyright violations, and identify articles needing improvement. It argues that categorically labeling all LLM development as unfair would contradict the broader purposes of copyright law, which should support, rather than stifle, technological innovation and community-driven knowledge. ## Broadening Amicus Participation Wikimedia’s filing joins a growing wave of amicus activity in the Southern District of New York, including submissions from Engine Advocacy, the Authors Alliance, and various publishing organizations. Judge Stein holds broad discretion to accept these briefs as the court continues to weigh summary-judgment arguments between the publishers and AI companies. ## Sources covering this story - https://www.unite.ai/wikimedia-foundation-seeks-to-file-amicus-brief-in-openai-copyright-case/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._