Mark Zuckerberg is due to be depositioned as part of a lawsuit filed by authors including comedian Sarah Silverman who claim Meta stole their content to train its artificial intelligence assistant.
U.S. District Judge Thomas Hixson on Tuesday denied Meta's motion to block Zuckerberg's deposition based on evidence presented by the authors that Zuckerberg was the “principal decision maker” for Meta's AI tools.
Silverman, along with authors Christopher Golden and Richard Kadry, filed a class action lawsuit against Meth last year.
The lawsuit alleges that Meta used stolen content from books via a “shadow library” website to train its AI language model, LLAMA, without consent or compensation.
The lawsuit is part of a growing number of copyright infringement lawsuits targeting AI chatbots.
The authors filed a similar lawsuit against OpenAI and its chatbot, ChatGPT, last year.
Mehta argued that Zuckerberg didn't have any unique knowledge about the company's AI initiatives that other employees couldn't get.
The case revolves around fair use issues, “which hinge primarily on the transformative nature of the AI models and their impact on the market for Plaintiffs' Books,” Mehta's lawyers wrote in court documents.
“These elements are not necessary to establish or defend voluminous depositions, let alone Mr. Zuckerberg's testimony,” the lawyers wrote.
But the court ruled on Tuesday that Zuckerberg was the “policy maker” of Meta's AI business.
The authors presented evidence of “his specific involvement in the company's AI initiatives” and “his direct oversight over Meta's AI products,” Judge Hixson wrote in his ruling.
The plaintiffs New York Times article Since April, it has been arguing that tech giants such as OpenAI, Google, and Meta are circumventing copyright laws to get ahead in the AI race.
According to the article, when OpenAI released ChatGPT in 2022, Zuckerberg felt he was lagging behind in the field and “immediately worked to match and surpass ChatGPT, making late-night calls to executives and engineers urging them to build a rival chatbot.”
Meta says the books in its AI training dataset are taken from publicly available datasets, but does not disclose their exact origins.
Mehta and lawyers for the plaintiffs did not immediately respond to requests for comment.
Another group of authors filed a similar lawsuit against AI startup Anthropic in August.
The class action lawsuit alleges that Anthropic used pirated copies of authors' works to train its chatbot, “Claude.”
