In a discovery order tied to a class action filed by authors such as Richard Kadrey and Sarah Silverman, as well as a separate suit from Entrepreneur Media, Magistrate Judge Thomas Hixon required Meta to identify the individuals who made, authorized, or approved the acquisition of copyrighted books through torrenting. The order also mandates the production of detailed command-history logs from every server used in the process.
Meta’s filing argued that its early AI-training data collection was managed through an informal system in which no senior employee formally decided to pirate books. The company contended that any “decision” was limited to staff members who technically performed the torrent downloads, thereby attempting to shield executives from direct responsibility.
Testimony from Meta employees, however, revealed discomfort with the practice. One worker noted that “torrenting from a corporate laptop doesn’t feel right,” while another explained that downloading certain datasets required “some director or VP level approval to use for training.” Internal messages also show a pause and later resumption of torrenting from the archive known as Anna’s Archive.
Unredacted communications released earlier in the year suggest that a request to download from LibGen was escalated to “MZ,” a reference to Mark Zuckerberg, indicating that senior approval was sought. Entrepreneur Media’s joint letter argued that this contradicts Meta’s claim that no one at the company decided to torrent any material, emphasizing repeated downloads between 2022 and 2024.
The judge’s order compels Meta to turn over “command history files for every server Meta used to torrent, including its virtual machines and AWS instances,” as previously reported by TorrentFreak. Those logs could provide the concrete evidence authors need to demonstrate that Meta not only downloaded but also seeded copyrighted works, and they may also reveal any mitigation steps the company employed.
Authors have asked Meta to list each natural person who made, authorized, or approved any decision to obtain a shadow-library dataset, along with dates of involvement and the specific actions taken. Meta countered by saying the request would require canvassing every employee who ever mentioned BitTorrent, but Judge Hixon rejected that narrow reading and instructed the company to comply with the broader identification demand.