Third Circuit Rules in Thomson Reuters v. ROSS on 29 September 2026: Training a Legal AI on 2,243 Westlaw Headnotes Was Not Fair Use

What the Third Circuit decided on 29 September 2026

In Thomson Reuters v. ROSS Intelligence, No. 25-2153, a Third Circuit panel of Judges Restrepo, Montgomery-Reeves and Bove affirmed partial summary judgment for Thomson Reuters. Judge Montgomery-Reeves wrote the precedential opinion. It holds that 2,243 Westlaw headnotes are original works and that ROSS's use of them to train its AI legal search engine was not fair use under 17 U.S.C. 107. ROSS had commissioned about 25,000 training memos built with the help of the headnotes, and it marketed the finished product as a substitute for Westlaw at comparable prices.

Why the intermediate-copying defense failed

ROSS argued that training is an intermediate step, protected like the code copying in Google v. Oracle, Sega and Connectix. The court read those cases as resting on necessity: copying was the only route to unprotected functional elements. ROSS had free access to the judicial opinions themselves and chose the headnotes because they were easier to use. The court also held that each headnote is a separate work, so every one copied was a whole work taken.

Which market the court protected

On the fourth factor the court found that a market for licensing headnotes as AI training data is rapidly developing, and that ROSS's copying usurped Thomson Reuters's chance to enter it. The opinion addresses a non-generative system and expressly leaves the generative-AI cases, including the pending OpenAI litigation in New York, to their own facts.