Magistrate Judge Thomas Hixson has ordered Meta Platforms to disclose the names of managers and executives who authorized the use of BitTorrent to acquire datasets for AI training. The ruling, issued on October 20, 2026, requires the company to identify the key decision-makers behind the downloading of materials from shadow libraries Anna’s Archive and LibGen.
The court order follows testimony from Meta employees who stated that supervisors and executives approved the torrenting activities. Meta previously argued that the process was handled by lower-level staff through team consensus without formal documentation, despite internal concerns regarding the legality of the practice.
The discovery order covers two lawsuits in the Northern District of California, including a class action by authors such as Richard Kadrey and Sarah Silverman, and a separate case from Entrepreneur Media. Both lawsuits focus on how Meta collected data to train its Llama large language models.
The legal dispute centers on the methods used to acquire files and the practice of seeding, particularly regarding the Scimag collection from LibGen. Meta allegedly downloaded over 81 TB of data from Anna’s Archive and additional tens of terabytes from LibGen between 2022 and 2024.
In September 2026, the court previously directed Meta to provide server command histories for the machines involved in the downloads. The plaintiffs argue that the company’s actions constitute willful and contributory copyright infringement, shifting the focus from the training process itself to the acquisition and distribution methods.
