A federal court has ruled in favor of Anthropic in a landmark lawsuit, affirming the company’s legal right to use copyrighted works to train its artificial intelligence models without obtaining permission from the authors. Judge William Alsup determined that Anthropic’s use of published books for training large language models (LLMs) is protected under the fair use doctrine. This decision represents one of the first significant judicial recognitions that AI firms may invoke fair use as a defense when leveraging copyrighted materials.
For many authors, artists, and publishers who have filed multiple lawsuits against major AI players including OpenAI, Meta, Midjourney, and Google, this ruling serves as a troubling precedent. While Judge Alsup’s decision is not binding on other federal courts, legal experts suggest it could strongly influence the way similar cases are interpreted moving forward, potentially favoring technology companies over individual creators.
At the heart of these lawsuits lies the notoriously complex fair use doctrine, whose framework has remained virtually unchanged since its inclusion in the Copyright Act of 1976—well before the rise of the internet and generative AI. Courts typically evaluate several key factors when determining fair use, including the purpose of the derivative work (such as educational use or parody), the commercial nature of the reproduction, and the degree to which the new work transforms the original material.
Companies such as Meta have previously contended that their training processes should qualify as fair use, but until now, judicial perspectives on these issues have remained uncertain. In the case titled Bartz v. Anthropic, the authors filing suit also challenged not only the legality of the company’s use of their content but also the methods Anthropic employed to acquire that content. According to court documents, Anthropic intended to create a comprehensive digital repository of all published books, but allegedly obtained millions of copyrighted works through downloads from unauthorized sources.
Although the court found Anthropic’s training procedures lawful under fair use doctrine, Judge Alsup specified that a separate trial must address the company’s alleged use of pirated material in assembling its extensive digital “library.” In his written decision, Alsup stated explicitly, “We will have a trial on the pirated copies used to create Anthropic’s central library and the resulting damages,” adding that subsequently purchasing legitimate copies would not eliminate previous liability for illegally sourced content but could potentially influence the calculation of statutory damages.
