Anthropic, the developer of the Claude AI assistant, has agreed to pay at least $1.5 billion to resolve a class-action lawsuit alleging the company used pirated copies of approximately 500,000 books to train its large language models. The settlement, announced September 5th, represents the largest copyright settlement in U.S. history and could fundamentally reshape how AI companies source training data.
The lawsuit, filed in the Northern District of California, accused Anthropic of downloading millions of copyrighted books from piracy sites including Library Genesis and Pirate Library Mirror without authorization. Authors represented by Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson will receive approximately $3,000 per infringed work under the settlement terms.
Judge William Alsup had previously ruled in June 2025 that while Anthropic’s use of lawfully purchased books for AI training qualified as fair use, the company’s reliance on pirated materials was “inherently, irredeemably infringing.” This mixed ruling set the stage for potentially devastating damages that could have reached tens of billions of dollars.
As part of the agreement, Anthropic must destroy the pirated libraries and any derivative copies within 30 days of final judgment. However, the settlement notably does not protect the company from future copyright claims or cover potential infringement from AI model outputs. The deal only releases claims for past conduct occurring before August 25, 2025.
The settlement comes amid broader tensions between AI developers and content creators over training data usage. While courts continue wrestling with fair use applications for lawfully obtained materials, this case establishes a clear precedent that using pirated content leaves companies vulnerable to massive liability. Legal experts suggest the resolution could encourage more licensing deals between AI companies and publishers, similar to how the music industry evolved after Napster.
The proposed settlement still requires court approval and could influence dozens of similar lawsuits currently pending against other major AI developers. Industry observers view this as a watershed moment that may force AI companies to reconsider their data sourcing strategies and invest more heavily in legitimate content licensing agreements.