Meta faces order to identify executives in AI training torrent approvals

Traders Union
All News
Business & Wealth
Companies
Meta AI training court order

Highlights

Magistrate Judge Thomas Hixon ordered Meta to identify all supervisors, managers, and executives who authorized or approved AI dataset torrent downloads, expanding beyond formal decision makers.
Meta must produce server command history files and logs for all torrent activity, including AWS instances, which could bolster plaintiffs' contributory infringement claims regarding Meta's handling and transfer of copyrighted works.
Court testimony cites 80.6 terabytes downloaded from LibGen and 35.7 terabytes from Anna's Archive, with leadership involvement reaching director- and vice president-level and mention of Mark Zuckerberg in escalation.

Court order expands disclosure demands

As first reported by Ars Technica, citing TorrentFreak, Magistrate Judge Thomas Hixon largely sided with authors and publisher Entrepreneur Media in ordering Meta to more clearly identify the people behind decisions to obtain or download shadow library datasets containing copyrighted works.

The order affects a class action brought by book authors including Richard Kadrey and Sarah Silverman, as well as a separate case from Entrepreneur Media. After testimony from Meta staff who downloaded torrents, the ruling opens the door to identifying a broader group of supervisors, managers, and executives whom employees said approved the activity.

Meta had argued in court filings that its early effort to compile large AI training datasets followed an "informal approach," with no one in authority making a formal decision. The company maintained that a decision existed only in the sense that staff members technically carried out the downloads, but Hixon said the plaintiffs' request was not limited to formal decisions and that Meta must do more to explain who made, authorized, or approved the actions.

Hixon also rejected Meta's position that collaborative discussions meant there was no single decision maker to identify. The request, he said, does not assume a lone approver and instead seeks all decision makers involved.

Litigation risk grows for Meta's AI data practices

Authors and publishers argue that Meta willfully induced infringement by downloading and distributing copyrighted works through torrenting. They say the scale of the downloads, including 80.6 terabytes from LibGen and 35.7 terabytes through Anna's Archive, suggests Meta could not have obtained the material quickly without also seeding files to other users, although Meta says it used mitigations intended to prevent seeding.

The discovery order also compels Meta to produce command history files for every server it used to torrent, including virtual machines and AWS instances. Those logs could help plaintiffs support contributory infringement claims by showing whether Meta facilitated transfers of their works and by shedding light on the company's claimed anti-seeding measures.

Court filings cited by plaintiffs describe internal unease among staff, including one employee who said torrenting from a corporate laptop "doesn't feel right" and another who testified that downloads used for training required director- or vice president-level approval. Plaintiffs also pointed to testimony that leadership paused and later resumed downloads from Anna's Archive, while an unsealed message chain allegedly showed a LibGen download decision came after "a prior escalation to MZ," a reference to Mark Zuckerberg.

Hixon further said Meta must answer questions about what happened to pirated works that were not used for AI training. Meta and counsel for Entrepreneur Media did not immediately respond to a request for comment.

In our earlier article on the Bank of England’s push for “bare metal” recovery planning, we explained why regulators want major banks to build independent fallback systems that can keep critical services running even if primary technology is compromised. The piece highlighted concerns that advances in AI are making cyber attacks larger and more complex, prompting supervisors to raise expectations for end-to-end operational resilience across the financial system.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.