Meta Cleared of Copyright Infringement in Llama Lawsuit

A judge dismisses a lawsuit alleging that Meta used books to train AI

Meta, the developer of Facebook and Instagram and the creator of its own AI, Meta AI, faced a lawsuit from several authors claiming that the company used copyrighted books to train its AI. However, Judge Vince Chhabria, overseeing the case, has dismissed the lawsuit.

ORDER DENYING THE PLAINTIFFS' MOTION FOR PARTIAL SUMMARY JUDGMENT AND GRANTING META'S CROSS-MOTION FOR PARTIAL SUMMARY JUDGMENT (PDF file) -

The lawsuit alleged that Meta CEO Zuckerberg authorized the Llama development team to use data sets containing copyrighted books and documents to train Llama. However, Meta argued that downloading copyrighted data was not a copyright violation.

Federal judge rules copyrighted books are fair use for AI training

Judge's Ruling

Judge Chhabria of the San Francisco Federal Court dismissed the plaintiffs' lawsuit on June 25, 2025, stating that the plaintiffs had made false allegations. The plaintiffs claimed that Meta's AI (Llama) could reproduce parts of their books, affecting their ability to license works as training data. However, Judge Chhabria deemed these claims unfounded.

The plaintiffs argued that Meta engaged in 'massive copyright infringement' by using authors' books from online repositories to train its generative AI, Llama. They contended that Meta should have purchased and licensed the literary works instead.

Meta's Defense

Meta defended its actions by citing US copyright law, stating that unauthorized copying and transformation of copyrighted material into something new constitute fair use. It maintained that the text generated by Llama differed significantly from the original books and that Llama does not output exact copies of the training materials.

Meta lawsuit over child safety on social sites escapes dismissal

Judge Chhabria concluded that the plaintiffs' arguments lacked merit, emphasizing that 'Llama is not capable of generating sufficient text for the plaintiffs' books.' He also noted that the plaintiffs did not have a legitimate claim to a market in licensing works as AI training data.

Implications

The ruling does not support Meta's overall argument regarding the lawfulness of using copyrighted material for training language models. However, it highlights the plaintiffs' failure to substantiate their claims of copyright infringement by Meta.

Meta dismissed concerns that adhering to traditional copyright laws would hinder technological progress, emphasizing the importance of compensating copyright holders if AI companies require copyrighted material for training purposes.

The plaintiffs' lawyers expressed disappointment with the ruling, stating that incorporating copyrighted works into AI models without permission or compensation typically violates the law. They disagreed with the court's decision in favor of Meta.

Meanwhile, Meta underscored the transformative benefits of open-source AI models and the critical legal framework of fair use in driving innovation and creativity in the field.

Legal Precedent

A similar case involving Anthropic, the developer of the AI chatbot Claude, resulted in a court ruling that training AI with legally acquired books, even without the author's permission, constitutes fair use and does not infringe on copyright.