A US federal judge has approved a $1.5 billion (€1.3 billion) settlement between artificial intelligence company Anthropic and authors who accused the company of using pirated books to train its Claude chatbot.
The agreement, approved on 20 July by US District Judge Araceli Martínez-Olguín in San Francisco, closes the largest copyright class action in US history and marks the first major settlement in a growing wave of lawsuits over how AI companies use copyrighted material to train their systems.
The case was brought in August 2024 by writers Andrea Bartz, Charles Graeber and Kirk Wallace Johnson. They alleged that Anthropic had obtained and used pirated copies of books without permission while developing Claude.
Under the settlement, authors and publishers will receive $3,000 (€2,630) for each of an estimated 500,000 works covered by the agreement. Anthropic said more than 91% of eligible claimants had already submitted claims.
Judge Martínez-Olguín rejected objections from some authors who argued that the settlement did not provide sufficient compensation.
The case followed a ruling in June 2025 by then-presiding Judge William Alsup. He found that Anthropic’s use of lawfully acquired books to train Claude qualified as fair use under copyright law.
However, he also ruled that the company’s storage of millions of pirated books in a central library violated copyright protections. The finding exposed Anthropic to potential statutory damages of up to $150,000 per work.
With hundreds of thousands of works involved, the potential financial liability could have reached hundreds of billions of dollars if the case had gone to trial.
Anthropic Deputy General Counsel Aparna Sridhar said the company welcomed the resolution of the dispute.
Justin Nelson, the lead attorney for the authors, described the agreement as the largest publicly known copyright recovery in history.
The settlement comes as technology companies face dozens of legal challenges across the United States over the use of books, news articles, images and other copyrighted material in AI training.
Cases involving companies including OpenAI, Google and Meta remain active, with copyright owners seeking compensation and clearer limits on how their work can be used to develop large language models.
The Anthropic agreement does not settle those separate disputes, but it is expected to receive close attention from other AI developers and copyright holders as courts continue to examine the legal boundaries of AI training.
