Appellate Court Weighs Copyright Case with "Sweeping Consequences" for US AI Leadership

PHILADELPHIA – A federal appeals court is currently considering a copyright dispute between defunct legal AI company ROSS Intelligence and legal research giant Thomson Reuters (Westlaw), a decision that ROSS Intelligence co-founder Andrew Arruda asserts will have "sweeping consequences" for artificial intelligence innovation in the United States. The case centers on whether ROSS Intelligence's use of Westlaw's "headnotes" to train its AI-powered legal search engine constitutes fair use and if these headnotes are copyrightable.

Andrew Arruda, whose company ROSS Intelligence aimed to make legal services more accessible through AI, emphasized the critical nature of the appellate court's ruling. "This is a chance for the appellate court to help secure the American lead in AI development — a lead in innovation critical to our economic success and national security," Arruda stated in a recent social media post. The appeal follows a lower court's revised opinion that found Westlaw's headnotes to be original and copyrightable, denying ROSS's fair use defense.

The legal battle began in 2020 when Thomson Reuters alleged that ROSS Intelligence illegally copied its copyrighted headnotes and key number system, which ROSS obtained through a third party after being denied direct access. ROSS Intelligence argues that its use of these headnotes was "quintessential fair use" for scientific progress and that the headnotes, often verbatim or near-verbatim quotes from uncopyrightable judicial opinions, should not be copyrightable. The company claims that its AI model requires "vast quantities" of data for training.

The Third Circuit Court of Appeals granted an interlocutory appeal to resolve two controlling questions of law: the required "creative spark" for originality in copyright and whether ROSS has a fair-use defense. This marks the first U.S. appeal to examine the intersection of copyright law and AI training data, setting a significant precedent for the burgeoning AI industry. The outcome could influence how AI developers access and utilize existing data for training, potentially impacting the pace and direction of AI innovation in the country.