Third Circuit: Westlaw headnotes copyrightable, AI training not fair use
A Third Circuit ruling reportedly holds Thomson Reuters' Westlaw headnotes copyrightable and finds ROSS Intelligence's use of them to train a competing legal-research tool was not fair use.
According to a Polsinelli alert summarizing the decision, the Third Circuit held that Westlaw's headnotes — the editorial paraphrases Thomson Reuters attaches to its judicial opinions — are protectable copyrighted works, and that ROSS Intelligence's ingestion of those headnotes to train a competing legal-research platform did not qualify as fair use. The two steps matter separately: the panel first declined to treat the headnotes as unprotectable facts or summaries, then assessed ROSS's training use against the statutory fair-use factors.
Why sophisticated counsel should care: headnotes and similar editorial annotations are a recurring input to legal AI systems, and a ruling that treats them as protectable output moves the risk surface for model builders from "no copyrightable material ingested" toward licensing and case-by-case factor analysis. It also sharpens the unresolved split over generative AI training claims, where most US litigation has so far settled or been diverted.
Concrete next steps: expect attention on whether Thomson Reuters seeks injunctive relief or a damages ruling, whether ROSS appeals, and whether parallel training disputes over Westlaw headnotes and bulk opinions in other circuits line up or diverge. Caveat: this cluster contains a single law-firm summary; the full opinion, reasoning and any separate opinions should be read before advising clients.