Meta chief executive Mark Zuckerberg will be deposed as part of a lawsuit from authors involving the companys artificial intelligence technology.

U.S. District Judge Thomas Hixson on Tuesday rejected Metas bid to bar the deposition of Zuckerberg, pointing to evidence indicating that hes the principal decision maker of the tech giants AI initiatives.

Authors Sarah Silverman, Richard Kadrey and Christopher Golden filed the proposed class action last year in California federal court. They accused Meta of copyright infringement for illegally downloading their books from shadow library websites and copying them without consent or compensation to train its AI system. The case is one of several from creators in a barrage of court challenges over the legality of the way large language models are trained and may help establish guardrails for the creation of technology.

In Tuesdays ruling, the court said that Zuckerberg is the policy setter for Metas generative AI branch and the development of the AI systems at issue in the case.

The authors have submitted evidence of his specific involvement in the companys AI initiatives, as well as his direct supervision of Metas AI products, Hixson wrote.

Meta had argued that Zuckerberg doesnt have unique knowledge of the companys operations warranting his deposition that cant be obtained from other employees or executives. It stressed that the primary issue in the case is fair use, which provides protection for the use of copyrighted material without a license or compensation to make a secondary work.

Fair use will largely turn on the transformative nature of the AI models and the alleged effect on the market for Plaintiffs books, lawyers for the company wrote in a court filing. The authors do not need a litany of depositions, let alone, Mr. Zuckerbergs deposition, to establish or defend these elements.

In opposition, the writers had argued that Zuckerberg personally issued directives guiding the development and sale of Metas AI products. They cited an article from The New York Times reporting that he immediately pushed to match and exceed ChatGPT, calling executives and engineers at all hours of the night to push them to develop a rival chatbot.

The court concluded that the authors made a solid case that this deposition is worth taking.

Meta doesnt disclose the origin of the books in its dataset used to train LLaMA. While the company has said that the works came from a publicly available dataset for large language models, it doesnt further describe the contents.

OpenAI faces an identical copyright infringement lawsuit from writers. On Tuesday, the two sides came to terms on protocols for inspection of the companys training data for review of whether copyrighted works were used to power its technology.