New YorkMeta stripped metadata from pirated books, publishers allege.
5 publishers including Hachette and Elsevier claim Zuckerberg personally cleared the plan to torrent millions of books and forgo licensing.
A 2025 case found fair use; the documented CEO instruction to skip licensing is the new suit’s distinguishing claim.
Sources: Variety
THE LIBGEN PROBLEM
Library Genesis is a shadow library hosting millions of pirated books and academic papers, originally seeded from Russian academic networks in the 2000s. It is the de facto training corpus for text models — comprehensive, well-formatted, and unambiguously infringing at the source.
FAIR USE AND THE FOUR FACTORS
US courts evaluate fair use through four factors from the 1976 Copyright Act: purpose of the use, nature of the work, amount taken, and effect on the market. Transformative purpose — turning books into model weights rather than reselling books — has carried recent AI rulings. But fair use analyzes the use, not the acquisition.
THE ACQUISITION QUESTION
The 2025 Bartz v. Anthropic ruling found training itself was fair use but separated out the acquisition: Anthropic's use of pirated copies was not protected, and the company settled the piracy claim while winning the training claim. Meta's case sharpens the same split — fair-use training does not launder infringing acquisition.
WHY METADATA STRIPPING MATTERS
Section 1202 of the DMCA makes it a separate violation to remove copyright management information — author, ISBN, rightsholder — when you know it will conceal infringement. Stripping metadata is independently actionable, separate from the underlying copying, with its own statutory damages of $2,500 to $25,000 per work.
THE CEO LIABILITY DOCTRINE
Corporate officers who personally direct or approve infringement can be held individually liable — the doctrine traces to the 1916 Dangler case and was reaffirmed for copyright in the Napster era. Documented instructions from a named executive convert a corporate suit into a personal one, with no shield from the corporate veil.
THE STATUTORY DAMAGES MATH
US copyright law provides $750 to $30,000 per work in statutory damages, rising to $150,000 per work for willful infringement. Across millions of books, the theoretical maximum exceeds any settlement in copyright history — which is why these cases almost always settle, and why the willfulness finding is the entire fight.