跳转到主要内容

此界面翻译为草稿,未经母语使用者审校。英文为源版本。 English

Kadrey v. Meta: judge grants Meta partial summary judgment on fair use for training Llama on 13 authors' books

2025年6月25日 · 自动从公开来源采集,未经人工审核。

这篇内容目前只有 英语 原文版本。

What happened

On 25 June 2025 Judge Vince Chhabria granted Meta's cross-motion for partial summary judgment in Kadrey et al v. Meta Platforms (N.D. Cal., Case No. 23-cv-03417-VC). Thirteen authors, mostly fiction writers, sued Meta for downloading their books from online shadow libraries and using them to train its Llama models. The plaintiffs argued Llama could reproduce snippets of their books and that Meta had diminished their ability to license their works as AI training data; the court found both arguments clear losers. The potentially winning argument — that Meta copied their works to build a product that will flood the market with similar works, causing market dilution — was barely argued and no evidence was presented on it, so the court had no choice but to grant summary judgment to Meta on the training claim.

Why it may matter to you

This is one of the first rulings on whether training a model on copyrighted books is fair use, and it turned on the evidence the plaintiffs did and did not present — not on a broad ruling that training is lawful.

What is not yet known

The ruling affects only these thirteen authors; it is not a class action, and the order says it does not establish that Meta's use of copyrighted materials to train its models is lawful in general.

来源