Stay informed with free updates
Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.
Meta’s use of millions of books to train its artificial intelligence models has been judged “fair” by a federal court on Wednesday, in a win for tech companies that use copyrighted materials to develop AI.
The case, brought by about a dozen authors, including Ta-Nehisi Coates and Richard Kadrey, challenged how the $1.4tn social media giant used a library of millions of online books, academic articles and comics to train its Llama AI models.
Meta’s use of these titles is protected under copyright law’s fair use provision, San Francisco district judge Vince Chhabria ruled. The Big Tech firm had argued that the works had been used to develop a transformative technology, which was fair “irrespective” of how it acquired the works.
This case is among dozens of legal battles working their way through the courts, as creators seek greater financial rights when their works are used to train AI models that may disrupt their livelihoods — while companies profit from the technology.
However, Chhabria warned that his decision reflected the authors’ failure to properly make their case.
“This ruling does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful,” he said. “It stands only for the proposition that these plaintiffs made the wrong arguments and failed to develop a record in support of the right one.”
It is the second victory in a week for tech groups that develop AI, after a federal judge on Monday ruled in favour of San Francisco start-up Anthropic in a similar case.
Anthropic had trained its Claude models on legally purchased physical books that were cut up and manually scanned, which the ruling said constituted “fair use”. However, the judge added that there would need to be a separate trial for claims that it pirated millions of books digitally for training.
The Meta case dealt with LibGen, a so-called online shadow library that hosts much of its content without permission from the rights holders.
Recommended
Chhabria suggested a “potentially winning argument” in the Meta case would be market dilution, referring to the damage caused to copyright holders by AI products that could “flood the market with endless amounts of images, songs, articles, books, and more”.
“People can prompt generative AI models to produce these outputs using a tiny fraction of the time and creativity that would otherwise be required,” Chhabria added. He warned AI could “dramatically undermine the incentive for human beings to create things the old-fashioned way”.
Meta and legal representatives for the authors did not immediately reply to requests for comment.