Reuters score a win on the fair use front lines

A US appeal court has released an important judgement which clarifies "fair use" in the favour of an original content creator.

by Rob Corbidge

Published: 14:54, 01 October 2026

​Another piece in the legal jigsaw currently under construction in American courts over publisher content taken without consent for use in a commercial AI product dropped this week, with Thomson Reuters emerging victorious in its appeal judgement against Ross Intelligence.

"The undisputed evidence reflects that Thomson Reuters’s materials possess a creative spark, and Ross aspired to be a direct competitor by using them for a highly similar purpose," wrote Judge Tamika Montgomery-Reeves of the 3rd US Circuit Court of Appeals. The full ruling is here.

Filed all the way back in 2020, before we'd even learned to despise OpenAI et al, this case is of importance to all publishers as it featured a defence of fair use by Ross Intelligence. This is a US legal concept many hapless Europeans, myself included, have had to familiarise ourselves with in this world-turned-upside-down content looting bonanza of recent years.

Personally, I'm with the school of thought that agrees such matters should be settled in court, thereby bringing a further body of case law into existence from which other conclusions can be drawn. Contrasted with those who believe in specific prior regulation of course, a route that seems to me to be less flexible, and unlikely to reflect real-world cases, and also likely only to restrain innovation.

With that said, to anyone with even the remotest idea of fairness somewhere in their soul, this case looked cut and dried in favour of TR. The very fact that so many of us have been watching for the ruling with some anxiety for the outcome tells you that, however, nothing can be taken for granted at the present. Even sense. This case wasn't directly about GenAI, but nevertheless, the actual mechanics of the case do shed legal light on the use of another party's content for the enrichment of those who did not originally create it. There were no fewer than six amicus briefs filed in the case, meaning six separate organisations, including the US News/Media Alliance and the Association of American Publishers, desired their formal interest and legal opinion on the case known to the court.

To recap, TR run a legal search engine called Westlaw, a specialist platform. In the case against Ross Intelligence filed by TR in 2020, it was claimed that Ross had copied proprietary and copyrighted content from Westlaw in order to train its own, and directly competing, AI-powered legal search engine. At the crux of the claim were Headnotes, these being the written summary of any given case, enabling users to absorb the salient points, facts and rulings of said given legal case. In short, these constitute the most important added-value content TR provides to the search engine customers.

Importantly, the court considered whether the use of Westlaw's content by Ross was transformative. A word that has come to mean a great deal in the AI-enabled age and one we all know the importance of. Well, they decided it wasn't.

To quote: "Ross uses Thomson Reuters’s headnotes to create a legal-research platform to help users find judicial opinion passages responsive to their searches. True, Ross did not use the headnotes in precisely the same fashion as Thomson Reuters. Ross took an intermediate step of using the headnotes to train an AI program, which arguably presents a slight degree of difference in use. But the undisputed evidence demonstrates that Ross used the headnotes to train an AI program for the benefit of its legal-research platform. So both Thomson Reuters and Ross use the headnotes to create and optimize a legal-research platform that helps users find responsive legal material. Thus, Ross’s use of the headnotes shares the same ultimate purpose as Thomson Reuters’s use, making Ross’s use minimally transformative, at best."

The situation in this case seems to be that Ross were using AI-woo-woo language to hide a good old-fashioned bit of content theft, hoping that the blinding light of new technology would hide their real intent. I feel like that about all AI businesses frankly, it's just the quality and level of woo-woo that changes.

That said, we're not getting giddy, this is just one in a series of such cases coming up and the first to fully conclude. It's not the laser-guided delineation between creation and copying that many in publishing would like to see, or are likely to see. Note that the appeals court's reasoning for its decision is currently sealed and will be unsealed next week, but roughly what the ruling means for publishers is that the most egregious examples of content theft can't be protected with a thin skin of "it's AI".

In the wider view, this is another example of publishers gaining legal ammunition. As Ricky Sutton points out this week, although there was disappointment that Judge Brinkema ordered only behavioural remedies in the Google ad tech case rather than structural ones, she has specifically left the door open for publishers, or groups thereof, to sue Google for the lost revenue they have suffered from its advertising domination. She's even given them the base figures to start their calculations from. It's a bit-by-bit fight, rather than one of sweeping victories.

There's another factor with the Reuters/Ross case, that in strict legal terms shouldn't be relevant, but in the real world it almost certainly is. For the judges and lawyers involved with this case, it actually concerns their profession. They will know what Westlaw is, and what it does. In theory, the ruling might be the same if it involved a database of budgerigar genetics, with breeding notes adding the value, instead of legal ones - but the fact this is a legal platform must bring it home to at least some legal minds that this is what the age of AI-enabled harvesting means, brought directly to their own profession's front door. 

Ross, incidentally, are no longer trading. What a shame.