Judge orders Meta to name executives who approved AI training piracy

A federal magistrate judge has ordered Meta to identify the executives who authorized torrenting pirated books from shadow libraries to train its AI models, rejecting the company's claim that no one with authority ever made that decision.

Magistrate Judge Thomas Hixon issued the discovery order on October 19 in litigation consolidated from a class action brought by book authors — including Richard Kadrey and Sarah Silverman — and a separate lawsuit from publisher Entrepreneur Media. TorrentFreak first reported the ruling. The order obliges Meta to clearly name key decision makers behind the downloads.

The stakes are considerable. Courts are still defining how copyright law applies to AI training data, and this case tests whether senior leadership can be held responsible when a company builds datasets from known pirate sources. Whatever names surface could shape both the legal exposure and the public record of how Meta assembled its early AI corpora.

What does the order actually require?

For weeks, Meta has blamed staff for downloading torrents, and it has resisted detailing the chain of command, especially for leadership involved in early AI training. Hixon largely sided with the plaintiffs' push to pierce that defense.

Following testimony from Meta employees who performed the downloads, the order could expose a wider group of supervisors, managers, and executives whom staffers said greenlit the piracy. The scope of what Hixon agreed to is described in his discovery order, which governs what Meta must now produce in the litigation.

In plain terms: Meta can no longer keep the identities of the people behind the decision out of the record.

What was Meta's defense?

Meta's argument, laid out in an October 6 court filing, is that its initial efforts to compile massive AI datasets were managed through an "informal approach" in which supposedly nobody with authority made any decisions.

Instead, Meta insisted that a "decision" was "made only in the sense" that some staffers "technically" downloaded torrents. That framing attempts to confine legal responsibility to the low-level employees who operated the software, not the leaders who directed the work.

The obvious weakness is structural: multi-terabyte dataset assembly for frontier AI training does not typically happen without budget, staffing, and executive awareness. Hixon's order signals the court found the "nobody decided" account insufficient.

In plain terms: a judge has declined to accept that piracy at this scale lacked a decision maker.

Why the shadow libraries matter

The downloads at issue came from Anna's Archive and LibGen, shadow libraries (unauthorized repositories of copyrighted books) that authors say Meta used at scale. Authors previously alleged Meta torrented over 81.7TB of pirated books for AI training, a figure that frames the size of the disputed corpus.

Kadrey and Silverman are named plaintiffs in the class action, alongside Entrepreneur Media's separate suit. Their cases are part of a broader wave of copyright litigation testing whether training AI on copyrighted works constitutes fair use.

If Meta's named executives include senior figures in its AI organization, the ruling converts a data-practices dispute into a question of individual and institutional accountability.

In plain terms: the case now aims at who ordered the downloads, not just that they happened.

What comes next

Meta must now comply with the discovery order and identify the decision makers, which will feed into subsequent litigation over liability. The company could challenge the order, but Hixon has already accepted the core premise that key actors can and should be named.

For the AI industry, the ruling is a warning that "informal" data practices offer no shield when courts start asking who authorized them.