Anthropic Agrees to $1.5bn Settlement Over Use of Pirated Books in AI Training

Artificial intelligence company Anthropic has agreed to pay at least $1.5 billion to settle a U.S. class action lawsuit accusing it of using pirated books to train its AI models, according to court documents filed Friday.

“This landmark settlement far surpasses any other known copyright recovery,” said plaintiffs’ attorney Justin Nelson. “It is the first of its kind in the AI era.”

The case was brought by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, who claimed Anthropic copied their books without permission to train Claude, the company’s AI chatbot and rival to ChatGPT.

In June, U.S. District Court Judge William Alsup ruled partly in Anthropic’s favor, stating that training AI models on books — whether purchased or pirated — was “fair use” because it so transformed the original works. “The technology at issue was among the most transformative many of us will see in our lifetimes,” Alsup wrote, likening AI training to how humans learn by reading.

However, Alsup rejected Anthropic’s attempt to claim blanket protection under fair use, ruling that its practice of downloading millions of pirated books to build a permanent digital library was not permissible.

Responding to the settlement, Anthropic’s deputy general counsel Aparna Sridhar said:

“We remain committed to developing safe AI systems that help people and organizations extend their capabilities, advance scientific discovery, and solve complex problems.”

San Francisco-based Anthropic, which recently raised $13 billion in a funding round valuing the startup at $183 billion, is competing with Google, OpenAI, Meta, and Microsoft in the fast-expanding generative AI industry.

According to the legal filing, the settlement covers around 500,000 books, translating to roughly $3,000 per work — four times the minimum statutory damages under U.S. copyright law. Under the deal, Anthropic must also destroy pirated files and any duplicates, though it may keep books it purchased legally.

“This settlement sends a strong message to the AI industry that there are serious consequences when they pirate authors’ works to train their AI, robbing those least able to afford it,” said Mary Rasenberger, CEO of the Authors Guild.

The deal, which still requires court approval, comes amid mounting legal pressure on AI firms over their training practices.

In a related case, a U.S. judge in June ruled in favor of Meta, finding that training its Llama AI model on copyrighted books was “transformative” and qualified as fair use.

Meanwhile, Apple is facing a new lawsuit from authors Grady Hendrix and Jennifer Roberson, who allege the company used pirated books to train its “Apple Intelligence” features. The complaint claims Apple sourced data from “shadow libraries” filled with pirated works.

Apple has yet to comment on the case, which seeks class action status.