Kadrey v Meta Platforms Inc: Difference between revisions

From AI Law Wiki
Jump to navigation Jump to search
(Migration export)
 
(Migration export)
 
(No difference)

Latest revision as of 02:34, 28 April 2026

Kadrey et al. v. Meta Platforms, Inc. (Case No. 3:23-cv-03417-VC) is a copyright infringement class action lawsuit filed on July 7, 2023, in the United States District Court for the Northern District of California before Judge Vince Chhabria, alleging Meta used pirated books to train its LLaMA AI models.[1][2]

Background

In July 2023, authors including Richard Kadrey, Sarah Silverman, and Christopher Golden sued Meta for copyright infringement, claiming the company used pirated books from shadow libraries (including Library Genesis and Z-Library) accessed via BitTorrent to train its LLaMA 1, 2, and 3 models without authorization or compensation.[3]

Plaintiffs later expanded to include additional authors (Junot Díaz, Ta-Nehisi Coates, Rachel Louise Snyder) and Entrepreneur Media, LLC.[4]

Claims

The complaint alleges:

  • Direct copyright infringement — copying plaintiffs' books for AI training
  • Vicarious copyright infringement — benefitting from infringement while having the right and ability to supervise
  • DMCA violations — removing copyright management information from works
  • Unfair competition under California law
  • Negligence under California law[4][5]

Key Rulings

June 25, 2025: Partial Summary Judgment on Fair Use

Judge Chhabria issued a significant order (Dkt. 598) granting Meta's cross-motion for partial summary judgment on fair use grounds, while denying plaintiffs' motion.[1][3]

The court analyzed the four fair use factors:

  • Factor 1 (Purpose and Character): Found Meta's use transformative despite its commercial nature and the sourcing from shadow libraries. No evidence showed Meta supported or encouraged illegal libraries.[4]
  • Factor 3 (Amount): Copying entire books was justified for training quality and effectiveness.[4]
  • Factor 4 (Market Effect): Plaintiffs failed to prove market harm, including inability to establish viable licensing markets for AI training or market dilution.[3][4]

The ruling was limited to these plaintiffs' claims and did not constitute a blanket approval of Meta's training practices.[3]

April 6, 2026: Fourth Amended Complaint Granted

Judge Chhabria reluctantly granted plaintiffs' motion to file a fourth amended complaint (Dkt. 658), adding a contributory copyright infringement claim related to Meta's alleged torrenting of plaintiffs' works from shadow libraries.[6]

Key aspects of the ruling:

  • The new claim alleges Meta knowingly induced or materially contributed to third-party infringement by torrenting and uploading copyrighted books, distinct from the direct infringement claims already adjudicated on fair use grounds.
  • Plaintiffs added expanded details on Meta's "non-training uses" of works from pirated shadow library sources.
  • Judge Chhabria strongly rebuked plaintiffs' counsel (Boies Schiller) for a "pattern" of blaming Meta for case delays, noting plaintiffs could have added these claims as early as November 2024.[6]
  • The court denied class discovery until named plaintiffs survive summary judgment on both distribution claims and the new contributory claims.[6]
  • Discovery is coordinated with the related Entrepreneur Media v. Meta case for shared discovery on contributory claims.[6]

Current Status (April 2026)

As of April 2026, the case is in its second phase, addressing distribution and uploading claims following the June 2025 fair use ruling on training use:

  • Phase 2 scheduling order: On March 27, 2026, Judge Chhabria issued a scheduling order for the distribution/torrenting phase. Expert discovery closes November 20, 2026. Plaintiffs' opening summary judgment brief is due December 11, 2026; Meta's opposition and cross-motion due January 11, 2027; hearing on dispositive motions set for February 25, 2027, in San Francisco.[7]
  • Contributory infringement claims: Plaintiffs' fourth amended complaint (filed April 6, 2026) adds claims for contributory infringement based on Meta's alleged torrenting and uploading activities. Judge Chhabria granted the motion but strongly rebuked plaintiffs' counsel for delay.[6]
  • Distribution and uploading claims: The court's June 2025 fair use ruling addressed training use only; distribution and uploading claims remain live issues, now scheduled for summary judgment in early 2027.[4]
  • No class certification yet: Judge Chhabria denied class discovery until named plaintiffs survive summary judgment on distribution and contributory claims.[6]
  • Related litigation: The case is consolidated with Chabon v. Meta, and the related Entrepreneur Media case advances contributory claims with coordinated discovery.[6][2]

The case is part of a broader wave of over 80 pending copyright lawsuits against AI companies testing the boundaries of AI training data usage.

Significance

Kadrey v. Meta is one of the most closely watched AI copyright cases, as Judge Chhabria's June 2025 fair use ruling represents one of the first judicial precedents on whether training AI models on copyrighted works constitutes fair use. The narrow scope of the ruling (limited to these plaintiffs' evidence) means the broader question remains unresolved and will be shaped by future rulings and appeals.[3][4]

See Also

References