Thomson Reuters v. Ross Intelligence: The case does not directly apply to generative AI, but the court’s reasoning on fair use could shape future AI copyright disputes.

AI companies cannot rely on fair use if they repurpose copyrighted content in ways that compete with original markets.

Analysis of Thomson Reuters v. Ross Intelligence (Final Verdict & Relation to Generative AI)

by ChatGPT-4o

Verdict Summary

The court ruled in favor of Thomson Reuters, granting partial summary judgment on direct copyright infringement against Ross Intelligence. The key rulings include:

1. Ross Infringed 2,243 Headnotes from Westlaw

   • The court found actual copying and substantial similarity between Westlaw's headnotes and Ross's Bulk Memo training data.

   • The only remaining factual issue is whether some of the copyrights have expired.

2. Ross’s Defenses Failed

   • Fair use: The court ruled against Ross, emphasizing that:

     • The copying was not transformative because Ross used Westlaw headnotes to train a competing legal research tool.

     • Ross’s use was commercial, directly competing with Westlaw.

     • The effect on the market for legal AI training data was significant.

   • Other defenses (innocent infringement, copyright misuse, merger, scenes à faire) were all rejected.

3. The Case Is NOT About Generative AI

   • The court explicitly distinguished Ross's AI from generative AI.

   • Ross's tool did not generate new legal content; it retrieved judicial opinions based on AI-enhanced search.

   • The court emphasized that the AI landscape is evolving, and its ruling applies to non-generative AI.

---

How This Case Relates to AI (and Generative AI)

1. Implications for AI Training Data & Copyright

• The ruling reinforces copyright protection over curated data used for AI training, setting a precedent that AI companies cannot freely scrape, use, or transform copyrighted material without permission.

• The court rejected fair use for AI training data when it competes with the original market (i.e., Westlaw’s legal research business).

2. Distinguishing Generative AI from AI Search

• Ross's AI is not generative AI (it does not create new legal text, only retrieves judicial opinions).

• This ruling does not directly address generative AI models like ChatGPT, Claude, or Google Gemini.

• However, the court’s reasoning on fair use and market harm could be used in future generative AI cases.

3. AI Companies Face Higher Copyright Risks

• AI companies cannot rely on fair use if they repurpose copyrighted content in ways that compete with original markets.

• Legal AI tools must either license training data or develop original datasets.

---

Key Takeaways

• Thomson Reuters won on key copyright claims, setting a precedent for protecting legal research materials.

• Ross Intelligence lost because its AI copied and repurposed copyrighted Westlaw headnotes to compete in the same market.

• AI developers (especially legal tech firms) now face stricter copyright limitations, particularly when training AI on proprietary datasets.

• The case does not directly apply to generative AI, but the court’s reasoning on fair use could shape future AI copyright disputes.