Updated
Updated · Ars Technica · Oct 20
Judge Orders Meta to Identify Executives Behind 116.3 Terabytes of AI Training Torrents
Updated
Updated · Ars Technica · Oct 20

Judge Orders Meta to Identify Executives Behind 116.3 Terabytes of AI Training Torrents

3 articles · Updated · Ars Technica · Oct 20

Summary

  • A discovery order in two copyright cases requires Meta to clearly identify the supervisors, managers and executives who approved downloading pirated books for AI training.
  • Judge Thomas Hixon acted after Meta spent weeks arguing no real decision-maker existed and that staff merely “technically” downloaded torrents under an “informal approach.”
  • Authors and Entrepreneur Media say the downloads—80.6 terabytes from LibGen and 35.7 terabytes via Anna’s Archive—necessarily involved illegal distribution because torrenting also seeds files to others.
  • Testimony and internal messages already pointed above rank-and-file staff: one employee said training downloads needed director- or VP-level approval, and an unsealed message said a LibGen decision followed “a prior escalation to MZ.”

Insights

Could Mark Zuckerberg face personal legal consequences for Meta's alleged use of pirated books to train its AI?
Will exposing the executives behind Meta's shadow library downloads trigger a massive financial reckoning for proprietary AI models?
If tech giants intentionally pirated data, will courts force them to completely destroy their billion-dollar AI systems?