Newly unsealed emails have exposed shocking details about Meta’s alleged use of pirated books to train its AI models, sparking a legal firestorm over potential copyright infringement.
Unsealed Emails Expose Meta’s Piracy Concerns
Internal emails from Meta employees reveal concerns about the legality of torrenting copyrighted books from sites like LibGen and Z-Library. One engineer, expressed discomfort with ‘torrenting from a corporate laptop’ and worried about using Meta IP addresses ‘to load through torrents pirate content.’
Hiding in ‘Stealth Mode’: Meta’s Alleged Seeding Tactics
Despite the concerns raised, authors allege that Meta chose to conceal its torrenting activities by operating in ‘stealth mode’ and modifying settings to minimize seeding. An internal message from a researcher discussed avoiding the ‘risk’ of being traced back as the ‘seeder/downloader’ from Facebook servers.
The Legal Minefield: Copyright Infringement Allegations
The authors’ court filings claim that Meta torrented ‘at least 81.7 terabytes of data across multiple shadow libraries,’ including 35.7 terabytes from Z-Library and LibGen, as well as 80.6 terabytes from LibGen alone. This alleged piracy has complicated Meta’s defense in the copyright infringement case.
Zuckerberg’s Involvement: Escalation and Decisions
Unredacted messages suggest that the ‘decision to use LibGen’ was made after ‘a prior escalation to MZ’ – potentially implicating Mark Zuckerberg’s involvement, despite his claims of having no role in the decision. Authors are now seeking to depose Meta staff again, alleging contradictions in previous testimony.
Seeding Expands Authors’ Distribution Theory
By allegedly seeding the pirated books through torrenting, authors argue that Meta has expanded their distribution theory beyond just claiming AI outputs unlawfully distributed their works. This could strengthen the direct copyright infringement claim against Meta.
The Magnitude of Meta’s Alleged Piracy Scheme
The authors’ filing describes the ‘magnitude of Meta’s unlawful torrenting scheme’ as ‘astonishing,’ stating that even much smaller acts of data piracy have resulted in criminal investigations. The sheer volume of copyrighted works allegedly pirated by Meta is staggering.
Meta’s Defense: Fair Use or Unlawful Conduct?
While Meta has maintained that its use of LibGen for AI training constitutes ‘fair use,’ the company now faces the challenge of addressing the seeding allegations. In a previous motion, Meta claimed authors failed to show instances of their books being distributed via torrent.
Depositions and Contradictions: The Next Legal Battle
With the new information revealed in the unsealed emails, authors argue that Meta staff involved in the LibGen decision must be deposed again, as the alleged facts ‘contradict prior deposition testimony.’ The legal battle is set to intensify as both sides grapple with the implications of Meta’s alleged torrenting activities.
