AI IP Year in Review – First Federal Ruling Rejects Fair Use Defense for AI Training Data in Copyright Dispute

In February 2025, a Delaware district court issued a much-awaited summary judgment decision in Thomson Reuters Enterprise Centre GmbH et al. v. Ross Intelligence Inc., No. 1:20-cv-613. This was the first of a series of district court decisions last year that began determining the contours of fair use analysis in copyright actions involving copyrighted works used to train artificial intelligence (AI) models.

As background, defendant Ross Intelligence developed AI-based legal research software using Westlaw’s copyrighted legal materials for training data. The court found Thomson Reuter’s copyright in its headnotes and Key Number system to be valid and granted partial summary judgment to Thomson Reuters on direct copyright infringement.

In its analysis, the court found that the fair use doctrine did not apply. For the first fair use factor (the purpose/character of the use), the court found Ross’s use was not transformative because it did not have a “further purpose or different character” from that of Thomson Reuters’ copyrighted work. Instead, Ross aimed to develop a competing legal research tool. The court also found the fourth fair use factor (the effect on the copyrighted work’s market) favored Thomson Reuters because Ross “meant to compete with Westlaw by developing a market substitute.” The court dismissed claims of public benefit, as legal opinions are freely available but Westlaw’s curated content is not.

Update

The Thomson Reuters decision rapidly became central to generative AI copyright litigation. The decision became highly cited, including by the court in Bartz v. Anthropic PBC. And although the court was careful to exclude generative AI from its analysis by distinguishing Ross’s AI tool, the court’s application of the Warhol framework could nevertheless apply to generative AI. Indeed, this case has since been applied in a matter involving an artificial intelligence tool that generates study materials. Barkley & Assocs., Inc. v. Quizlet, Inc., No. 2:24-CV-05964-WLH-E, 2025 WL 2946993, at *6 (C.D. Cal. Sept. 11, 2025)

The ruling strengthened plaintiffs’ arguments that AI training isn’t transformative use and emphasized market harm, particularly the impact on emerging AI training data licensing markets. However, the court certified this case for interlocutory appeal, noting that “[b]ecause appellate resolution of these questions would change the shape of trial—and possibly avoid a copyright trial altogether—I stay this case pending the Third Circuit’s response.”

Takeaways

• Transformative use—the go-to defense for AI defendants—may not hold much water when the defendant’s product competes with the copyright holder. And intermediate copying may be relegated to specific factual scenarios.
• The fourth fair use factor could be met by a potential market for AI training data. This increases an AI defendant’s potential to fail the fourth factor when copyrighted material is part of its training data.

Read the original client alert, Fair-Use Whiplash: Thomson Reuters v. Ross Intelligence Revisits Fair Use for Artificial Intelligence.

This article appeared in the 2025 AI Intellectual Property: Analysis & Trends Year in Review report.