Meta CEO Mark Zuckerberg allegedly authorized the use of a dataset containing pirated ebooks for training the company's Llama AI models, according to a copyright lawsuit against Meta. The lawsuit claims that Meta's actions, including torrenting the dataset LibGen and stripping copyright information, constitute copyright infringement. Plaintiffs, including notable authors, argue that Meta's reliance on fair use is flawed. The case highlights ongoing disputes over AI training practices among tech giants.