Here’s a brief update and a ChatGPT-generated (we couldn’t help ourselves) summary of the recent, high-profile Reuters v. Ross lawsuit over the copyright dispute in which Thomson Reuters sued competitor Ross Intelligence for “scraping” legal headnotes to create its own content.
While this case is not exactly analogous to the recent lawsuits filed against AI music creation companies Suno and Udio, in our non-legal, non-expert opinion, it may portend well for the future outcomes of music-related infringement suits. This case was not about generative AI, but the way in which this court treated the fair-use argument may bode well for similar arguments regarding what fair use is, as it pertains to the wholesale scraping of music to train the AI that’s used by Suno, Udio, and other such companies. Read on, and stay informed!
In Reuters v. Ross, the United States District Court for the District of Delaware revisited its earlier summary judgment decision regarding a high‐profile copyright dispute over legal research materials. Thomson Reuters, owner of Westlaw—a comprehensive legal research platform featuring case law, statutes, and editorial headnotes—sued Ross Intelligence, a competitor that built an AI legal research tool. Ross had trained its tool using “Bulk Memos,” compilations of legal questions created with guidance from Westlaw headnotes, which Thomson Reuters contends are copyrighted due to the creative editorial process used to distill key legal points from lengthy judicial opinions.
The court granted partial summary judgment for Thomson Reuters on two critical fronts. First, on the issue of direct copyright infringement, the judge found that Ross had copied 2,243 headnotes. These headnotes, though derived from uncopyrightable judicial opinions, are sufficiently original because they involve a creative selection, arrangement, and condensation process analogous to sculpting raw material into a unique work. The court ruled that even when portions of the text are taken verbatim, the editorial choices imbue the headnotes with the “creative spark” required for copyright protection. Second, regarding fair use defenses, the court sided with Thomson Reuters, rejecting Ross’s counterarguments including claims of innocent infringement and copyright misuse.
Additionally, the court’s analysis extended to Westlaw’s Key Number System, a proprietary taxonomy used to organize legal information. Although questions remain as to whether all elements of the system were used by Ross, the judge maintained that the organization and classification decisions, even if partly automated, satisfy the minimal originality standard.
Some issues remain unresolved for trial, such as determining which headnotes are still covered by valid copyrights. However, the overall outcome significantly favors Thomson Reuters, establishing that Ross’s use of Westlaw-derived materials to train its AI tool constituted clear copyright infringement. This ruling underscores the legal protection afforded to creative editorial compilations, even when built upon public domain judicial opinions, and clarifies the limited scope of defenses like innocent infringement in cases involving clearly marked copyrighted material.
If you’d like to read the court’s actual ruling, you can do so by clicking here.