The Thomson Reuters v. Ross Intelligence Inc. case has had some significant movement this week - the judge granted partial summary judgment in favor of Thomson Reuters.  The case circles around the issue of whether it is a fair use when an artificial intelligence (AI) company uses another party's works (here, Thomson Reuters) as training data for its AI and machine learning. Notably, this case is about training AI, not generative AI, but the implications will likely be significant regardless.

Copyright fair use is analyzed using four factors in a balancing test - (1) the purpose and character of the use, (2) the nature of the copyrighted work, (3) the amount and substantiality of the portion used, and (4) the effect on the potential market of the work.  Here, the court ruled that on the balance, these factors weighed in favor of Thomson Reuters, concluding that using another party's works as training data for AI is not a fair use.

Fair use is the key copyright doctrine at play in these AI-copyright cases. To see a federal judge in Delaware conclude that training AI on another's work is not a fair use is surely sending alarm bells off in tech companies everywhere. While this is only one judge's ruling and each AI-copyright case will have different facts that may necessitate a different outcome, this case serves as a significant data point in what will become a tapestry of court decisions.