Federal Judge Rules AI Training on Copyrighted Content Infringes Copyright, Rejects Fair Use Defense

On Feb. 11, a federal judge granted summary judgment to a copyright holder, determining that a software company’s use of copyrighted content to train an AI legal research tool constituted direct copyright infringement and rejecting the AI tool company’s fair-use defense. Judge Bibas’ opinion departed from the seminal “intermediate copying” cases in Google and Sega and applied the Supreme Court’s “transformative use” analysis from Warhol.
Although it is expected that the defendant software company will appeal the grant of summary judgment, this opinion is one of the first to evaluate the use of copyrighted content to train and build AI tools. Other courts are likely to look at its reasoning. Interestingly, today’s opinion revised an earlier opinion, largely denying the plaintiff copyright holder’s earlier motion for summary judgment in the same case. Emphasizing the quickly changing technological landscape, the court’s opinion explicitly noted that it did not consider whether the result would differ if the AI product were generative AI.
Taft’s Technology and Artificial Intelligence and Intellectual Property teams continue to advise both copyright holders and users of copyrighted material as they do business in training and building AI tools and otherwise at the cutting edge of technology.
In This Article
You May Also Like
AI Blind Spot: Is the Board Exposing the Company to Uninsured AI Risks? AI-Powered Fraud: Immediate Action Steps to Protect Companies From Next-Generation Payment Scams