Anthropic AI Settlement Pays Publishers Millions for Copyright Infringement

A U.S. federal judge has given final approval to a $1.5 billion copyright settlement that requires artificial-intelligence startup Anthropic to compensate thousands of authors whose books were used to train its Claude chatbot. The ruling states that the settlement offers “meaningful relief” to the affected creators and publishers.

Associated Press

The case began in 2024 when novelist Andrea Bartz and two other writers filed a class-action lawsuit alleging that Anthropic had obtained millions of books from pirate websites and used them without permission to train its AI models. Earlier rulings found that training AI on copyrighted works was not illegal per se, but that Anthropic’s acquisition of the books was wrongful. The settlement follows dozens of similar copyright actions filed by authors and news outlets across the United States.

Associated Press

Bloomsbury Publishing, the London-based home of the “Harry Potter” series, listed 14,087 titles in the settlement and expects roughly $3,000 per title. After a 10 % deduction for attorney fees and expenses, Bloomsbury and the associated authors are projected to receive about $19 million (£14 million) in installments, potentially beginning in the second half of the current fiscal year. Overall, about 91 % of the more than 482,000 works covered by the case have been claimed, making this the largest known copyright recovery in history.

The Guardian

District Judge Araceli Martínez-Olguín described the settlement as providing meaningful relief to the class. Anthropic’s deputy general counsel, Aparna Sridhar, expressed satisfaction that the majority of authors and publishers have claimed their share and indicated the company looks forward to concluding the matter. Plaintiff attorney Justin Nelson emphasized the historic size of the recovery and pledged prompt distribution to class members. Earlier, U.S. District Judge William Alsup had issued a preliminary approval before retiring.

Associated Press

The settlement underscores growing pressure on AI developers to obtain licenses or pay for the copyrighted material used in model training. Creative professionals worldwide have called for clear permission requirements, while some U.S. AI firms continue to argue that fair-use defenses apply. The outcome may prompt additional licensing agreements, such as Bloomsbury’s recent AI-licensing deal that lets authors opt in for royalties when their works are used for training. The case signals a shift toward monetizing the data that powers generative-AI systems and could shape future litigation and policy discussions.

The Guardian