CATEGORIES

Court Rejects Fair Use Defense in AI Copyright Case

2.18.25

In a landmark ruling, U.S. Circuit Judge Stephanos Bibas of the U.S. District Court for the District of Delaware has rejected Ross Intelligence’s fair use defense in its copyright dispute with Thomson Reuters. The case, Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc., No. 20-613 (D. Del. 2025), marks the first judicial decision in the United States addressing the fair use doctrine in the context of artificial intelligence training.

Background

Ross Intelligence, a now-defunct legal research company, was accused of using Thomson Reuters’ copyrighted Westlaw headnotes to train its AI-based legal research platform. The headnotes, which summarize legal principles from judicial opinions, are created by Westlaw’s attorney-editors and serve as a key feature of the platform. Ross had previously attempted to license the headnotes but, upon being denied, allegedly resorted to scraping Westlaw’s content to develop its AI model.

Key Findings of the Court

Judge Bibas’ ruling largely granted summary judgment in favor of Thomson Reuters, concluding that Ross’ use of the copyrighted material did not qualify as fair use under the four-factor analysis established by 17 U.S.C. § 107.

  1. Purpose and Character of the Use
    • The court found Ross’ use to be commercial and non-transformative. Because Ross used the headnotes to create a competing legal research tool rather than adding new meaning or expression, the first fair use factor weighed against it.
  2. Nature of the Copyrighted Work
    • While legal headnotes involve factual elements, the court determined they also contain substantial editorial judgment, making them more creative than purely factual compilations. This factor weighed slightly against fair use.
  3. Amount and Substantiality of the Portion Used
    • Ross copied a substantial number of headnotes to train its AI model. Even though the output did not reproduce the headnotes verbatim, the court found the copying extensive enough to weigh against fair use.
  4. Effect on the Market
    • The most critical factor in Judge Bibas’ decision was the potential harm to the market for licensing AI training data. The court emphasized that even if Thomson Reuters had not yet licensed its headnotes for AI training, the unauthorized use by Ross Intelligence undermined a potential market, weighing heavily against fair use.

Significance and Implications

This ruling is a pivotal moment in the ongoing debate over AI and copyright law. While the decision does not directly address generative AI, its reasoning is already being cited in cases involving companies such as OpenAI and Meta Platforms. The ruling underscores the importance of obtaining proper licenses when using copyrighted materials for AI training and signals that courts may take a stricter stance on fair use claims in AI-related cases. As AI development continues to push legal boundaries, this case sets a precedent that intellectual property owners will likely rely on to challenge unauthorized use of their works in AI training. For legal practitioners and AI developers, the decision highlights the necessity of navigating copyright law carefully to mitigate potential liability.

 

 

Seth R. Ogden, Ph.D. | Shareholder