A US Circuit judge sitting in US District Court in Delaware rejected a claim by Ross Intelligence Inc. that its use of Westlaw content owned by Thomson Reuters represented fair use. It represents the first time that fair use doctrine was applied to the use of copyrighted works as training data for artificial intelligence.
Companies that include Facebook parent Meta Platforms, Microsoft, OpenAI and other developers of generative AI technologies have used fair use doctrine as a defense in copyright cases brought by copyright owners ranging from individual creators to large media companies like The New York Times and Thomson Reuters. The AI technology being applied by Ross was its legal search engine, not a generative AI platform that creates new material.
The court compared the Thomson Reuters content in its original form against the content found in Ross’ platform and found a strong similarity. Ross had claimed that its use of this content was innocent, which the judge rejected because the language between the two was very similar.
This February decision reverses a 2023 decision by the same court, based on a new interpretation of four factors in the case. Two of those factors favored the plaintiff, and two favored the defendant. The purpose of use and the value of the use were weighted more heavily.
Further reading
Thomson Reuters wins AI copyright ‘fair use’ruling against one-time competitor. Article. February 11, 2025. by Black Brittain. Reuters
Client Alert: Court decides that use of copyrighted works in AI training is not fair use: Thomson Reuters Enterprise Centre GmbH v. Ross Intelligence Inc. Article. February 12, 2025. by Steve Englund, Zach Marino. Jenner & Block.
Why it matters
According to the above-linked analysis by Jenner & Block, “Perhaps the most interesting takeaways here are the court’s willingness to consider the totality of Ross’s product plans, and its assessment that those plans involved a purpose similar to and competitive with Westlaw headnotes, rather than focusing narrowly on the immediate purpose of Ross’s copying. This approach could be influential to courts considering fair use issues involving generative AI systems.”
