主要出版商起诉Meta，指控Llama训练侵权
macmillan、mcgraw-hill、cengage、elsevier和hachette对meta提起集体诉讼，指控该公司非法从libgen和anna’s archive盗版书籍以训练其llama模型。投诉称llama能逐字输出受版权保护的教科书内容，如stewart的微积分。meta以合理使用为由反驳，并声称将积极应诉。
Five major publishers—Macmillan, McGraw-Hill, Cengage, Elsevier, and Hachette—alongside author Scott Turow, filed a class action lawsuit against Meta in federal court. The complaint alleges Meta engaged in systematic copyright infringement by training its Llama models on pirated content sourced from sites like LibGen, Anna's Archive, and Sci-Hub. The publishers cite internal company communications and claim Llama can reproduce verbatim passages from protected works, such as Cengage's *Calculus: Early Transcendentals*.
The plaintiffs seek damages and an injunction to halt Meta's use of their copyrighted materials, alongside a court order requiring Meta to disclose the specific datasets used for training. Meta spokesperson Dave Arnold defended the company's practices, stating courts have found AI training on copyrighted material qualifies as fair use, and confirmed Meta intends to fight the lawsuit aggressively.