Meta's AI Training Data Scandal: Leaked Emails Expose Torrenting of 81.7TB of Pirated Material

Meta is facing a class-action lawsuit for copyright infringement. Newly unsealed court documents reveal internal communications showing Meta employees used at least 81.7TB of illegally torrented data from shadow libraries like LibGen to train their AI models. Emails reveal employees voiced ethical concerns and even used VPNs to mask their IP addresses while torrenting and seeding copyrighted material. This contradicts previous depositions, bolstering the plaintiff's case and potentially significantly damaging Meta's defense.