Public Beta · Not legal advice
← Interactive map

United States · US-009

Kadrey v Meta Platforms

HELDU.S. District Court for the Northern District of California3:23-cv-03417-VCTraining claim decided; other claims continued
TrainingTraining Data AcquisitionFair UseMarket Harm / SubstitutionSecondary / Contributory Liability

Key holding or allegation

Meta obtained summary judgment on the named plaintiffs' training claim because the evidentiary record did not establish sufficient market harm, despite the court recognising that generative-AI market dilution can be legally important.

Why it matters

Major U.S. merits ruling warning against categorical treatment of AI training while showing the centrality of market-harm evidence.

Facts

Authors alleged Meta used copyrighted books, including shadow-library materials, to train Llama and engaged in related distribution/torrent conduct.

Issues before the court

Whether the challenged training was fair use and whether market harm/dilution was sufficiently evidenced; separate acquisition/distribution theories continued.

What the court did not decide

Does not hold that all AI training is fair use; later torrent/distribution/contributory theories remained live.

Procedural timeline

2025-06-25

U.S. District Court for the Northern District of California
Summary judgment: Meta prevailed on the named plaintiffs' training claim because the evidentiary record did not establish sufficient market harm, while the court recognised that generative-AI market dilution can matter to fair use.

HELD