Thomson Reuters scored a victory defending its intellectual property in the first AI model training case to produce a substantive legal judgment. U.S. District Court of Delaware Judge Stephanos Bibas on Tuesday issued a partial summary judgment for Westlaw parent Thomson Reuters in its copyright infringement case against Ross Intelligence. The court found that after Thomson Reuters refused Ross’ offer to license Westlaw material the startup hired a third-party to procedurally reconstitute the material, resulting in infringement. Ross defenses, including fair use, “all fail,” says the court.

“Similar lawsuits against OpenAI, Microsoft, and other AI giants are currently winding their way through the courts, and they could come down to similar questions about whether or not the AI tools can claim a ‘fair use’ defense of using copyrighted material,” writes The Verge.

Wired calls the battle “the first major AI copyright case in the U.S.” and says this early decision “has big implications for the battle between generative AI companies and rights holders.”

In his 23-page opinion Bibas makes clear the case before him was for “actual copying” for the purposes of model training, “not generative AI” (in that the Ross platform didn’t craft its own responses but returned previously written material).

The case hinges on how words and phrasing used in Westlaw’s human-authored summaries, known as “headnotes,” link to specific cases and citations via “Key Number,” Westlaw’s proprietary cross-reference system.

The Register writes that “at least 38 AI-related copyright claims are pending before U.S. courts” and cites a legal expert who believes “courts hearing generative AI cases will consider judge Bibas’s reasoning.” Bibas’ opinion may be influential because he ruled as a matter of law on a fundamental issue — that unauthorized data scraping of copyrighted material is illegal.

Since his judgment was partial, some issues remain for trial. While most of the legal writing that Westlaw sells is not copyrightable, Bibas ruled that its headnotes and Key Number system are and that 2,243 headnotes were illegally appropriated. An additional 5,367 headnotes and question of whether the Key Number system was infringed was left for a jury.

Despite those loose ends, “the opinion seems to be striking victory for content owners in their fight against the AI onslaught,” suggests The National Law Review, writing that it “will be worth watching” to see if the ruling can withstand an appeal. Damages will be determined at trial or by settlement.

Ross, which launched in 2014, “shut down in 2021, calling the lawsuit ‘spurious’ but saying it was unable to raise enough funding to keep going while caught up in a legal battle,” according to The Verge.