Meta accused of using pirated books to train his AI

A group of authors, including writer Ta-Nehisi Coates and comedian Sarah Silverman, are suing Meta, accusing the company of knowingly using pirated books to train its artificial intelligence models.

According to court documents made public this week in California, Mark Zuckerberg personally approved the use of copyrighted content to train its artificial intelligence.

At the heart of this controversy is LibGen, a database containing approximately 32 terabytes of digitized booksconsidered a “ shadow library » of Russian origin. According to the plaintiffs, internal communications reveal that Zuckerberg approved the use of LibGen despite warnings from his management team about the pirated nature of the content.

Also read – Microsoft AI boss says public content on the Internet is free to use, and that suits him just fine

Mark Zuckerberg would have approved this controversial practice

The court documents notably cite a memo referencing Zuckerberg's initials (“MZ”), stating that ” after ascent to MZ “, Meta's AI team has received permission to operate LibGen. The company's engineers even reportedly discussed access to this data, while expressing reservations about downloading torrents from professional computers.

This new legal action follows a first complaint filed in 2023, rejected by federal judge Vince Chhabria. The authors say this new evidence not only strengthens their original accusations of copyright infringement, but could also justify the addition of a computer fraud charge.

Context is particularly sensitive for Meta, which uses this data to train Llama, its language model powering its chatbots. Last year, a federal court in New York had already ordered the anonymous operators of LibGen to pay $30 million in damages to a group of publishers for copyright infringement.

This case is part of a broader debate on the use of protected content for training generative AI models. Creative professionals and publishers are warning of the risks that these practices pose to their livelihoods and their economic models.

Although Justice Chhabria agreed to the filing of an amended complaint, he expressed dhe doubts about the chances of success of the fraud charges. Meta has yet to publicly respond to these new allegations.