Anthropic has agreed to pay at least $1.5 billion to settle a United States class action lawsuit over claims it used pirated books to train its artificial intelligence models, according to court documents filed on Friday.
Plaintiffs’ lawyer Justin Nelson described the deal as “a landmark settlement” and “the first of its kind in the AI era,” noting that it far exceeds any previous copyright recovery.
In June, US District Court Judge William Alsup ruled that training Anthropic’s Claude models on books — whether purchased or pirated — was sufficiently transformative to qualify as “fair use.” However, he rejected Anthropic’s bid for blanket protection, ruling that building a permanent library of pirated works was unlawful.
The agreement covers around 500,000 books, with authors receiving roughly $3,000 per work — four times the minimum statutory damages under US law. Anthropic will also destroy all pirated files and copies, retaining only legally obtained works.
Mary Rasenberger, CEO of the Authors Guild, said the deal sends “a strong message to the AI industry” about the consequences of misusing authors’ work.
Anthropic, valued at $183 billion after a $13 billion funding round this week, competes with OpenAI, Google, Meta, and Microsoft in the fast-growing generative AI sector.
Meanwhile, Apple faces a fresh lawsuit from authors Grady Hendrix and Jennifer Roberson, who allege the company used pirated books to train its new “Apple Intelligence” features. The case seeks class action status.