In a significant decision entered on February 11th, 2025 in the United States District Court for the District of Delaware (Case No. 20-cv-0613-SB) the Court granted partial summary judgment in a copyright case against a legal AI company, Ross Intelligence, and in favor of Thomson Reuters, owner of Westlaw. As set forth in the Court’s memorandum opinion, Thomson’s copyright claims were based on Ross’s use of thousands of “Bulk Memos” that were created by a second company LegalEase. The Court found that Ross paid LegalEase to create the “Bulk Memos” using Westlaw headnotes and then used the “Bulk Memos” to train its AI. And the Court also found that Ross had previously tried to license the headnotes from Thomson, but Thomson refused.

In parsing through many issues, the Court ruled that over 2,000 of the “Bulk Memos” that were ingested by Ross were so obviously similar to the headnotes that no reasonable jury could find anything other than copyright infringement. The direct copyright infringement decision is notable in that it involved intermediate copying only, meaning that the headnotes do not appear in the end-product given by Ross to the consumer, but rather the copying occurred at an intermediate step, where Ross through LegalEase trained its AI on the copied content, in essence “turn[ing] the headnotes into numerical data about the relationships among legal words to feed into its AI.” The Court also considered but rejected a challenge based on claimed thin copyright protection for the headnotes, in other words that Westlaw headnotes should not have significant copyright protection because they are not sufficiently creative or original, as they are based on summarizing parts of the caselaw opinions themselves, which are not copyrightable.

The Court also found against Ross on all of its affirmative defenses, including fair use. The fair use ruling is significant because, although the Court found the four fair use factors to be split two for each side, the Court still found that the scales tipped in Thomson’s favor as the use was not transformative and the effect on the market as a substitute was significant. In finding against fair use, the Court directly addressed that the copying here was intermediate only, but distinguished and dismissed a number of cases that found intermediate copying as fair use, as being computer-programming related copying cases or cases where the copying was necessary for competitors to innovate (which the Court did to find to be the case for Ross). While citing to the case at various points, it bears noting that the Court did not directly grapple with how its fair use analysis squares with the Second Circuit’s fair use analysis in Authors Guild v. Google, Inc., 804 F.3d 202, 220 (2d Cir. 2015). On the other hand, in distinguishing prior caselaw, the Court here made clear that it believed its decision “fits more neatly into the newer framework advanced by” the Supreme Court recently in Andy Warhol Found. for the Visual Arts, Inc. v. Goldsmith, 598 U.S. 508, 529–31 (2023).