Meta CEO Mark Zuckerberg authorized the use of LibGen, a dataset of pirated e-books and articles, to train the company’s Llama AI models, according to new court documents. The revelation comes from a copyright lawsuit filing reported by Kyle Wiggers in TechCrunch. The lawsuit, Kadrey v. Meta, includes plaintiffs Sarah Silverman and Ta-Nehisi Coates. Court documents show Meta employees referred to LibGen as “known pirated” material, and engineers allegedly wrote scripts to remove copyright information from the training data. Meta’s AI team received approval after “escalation to MZ” despite internal concerns about regulatory implications. The company reportedly chose this approach over negotiating licenses, believing fair use would serve as a legal defense. The case currently only applies to Meta’s earliest Llama models, and the court has yet to make a final ruling.