On June 25, 2025, Meta secured a major legal victory in a high-profile U.S. copyright lawsuit accusing its AI models of unlawfully scraping authors’ works. A federal judge in California dismissed the proposed class-action lawsuit because the claimants failed to prove Meta’s AI systems directly copied their books.
In their legal action, Sarah Silverman, Richard Kadrey, and other authors alleged that Meta’s AI models had violated copyright laws by using their works for training without authorisation. The plaintiffs contended that Meta built its AI datasets by scraping “shadow libraries” that kept pirated copies of their works.
The judge dismissed the case, emphasising that there was no apparent similarity between the authors’ original works and Meta’s AI outputs. Although copyrighted texts may have been used in the training process, the court recognised that the AI’s outputs were transformative and were not identical to the protected content.
Meta claimed that its publicly accessible data, even from controversial sources, qualified as fair use, which recent decisions about artificial intelligence have reinforced. This company further stated that its models generate original content instead of repeating exact copies.
The outcome demonstrates the continuing conflict between copyright protection and AI innovation, as Many courts are increasingly supporting tech companies under the fair use doctrine. Since it allows AI companies to train models on copyrighted content without paying for it as long as the outputs aren’t direct copies, legal professionals suggest that this could set a dangerous precedent for individuals who create content and art.
But the battle is not quite over. Lawmakers in the US and the EU are debating new legislation to close the gap, while advocacy groups caution that unrestricted AI training could discourage artistic production. Upcoming laws may mandate openness regarding the sources of training data, which could compel businesses like Meta to reveal—or even license—copyrighted content.
The authors may appeal the decision for now, while the government faces mounting pressure to update copyright laws for the AI era. For now, Meta’s win strengthens the tech industry’s stance that AI training is fair use—a precedent with global ramifications.