Anthropic settlement draws nearly 120,000 author claims

Nearly 120,000 authors and other copyright holders have filed claims to share in Anthropic’s US$1.5 billion class-action settlement over the unauthorized use of books in AI training, with claims covering 91% of more than 480,000 works, court filings in California show.

Anthropic, a US AI startup backed by Amazon and Alphabet, is due at a May 14 hearing, where a judge will decide whether to give final approval to what has been described as the largest settlement in a US copyright case.

The case began after authors said Anthropic used pirated books to train Claude. A judge later found the training was fair use, but said the company still violated their rights by storing more than 7 million pirated books in a central library that would not necessarily be used for AI training.

🔗 Source: Reuters

🧠 Food for thought

Implications, context, and why it matters.

The lawsuit turns on a legal distinction

The settlement focuses on how training data was obtained, rather than the legality of AI training itself 1.
In a June 2025 ruling, Judge William Alsup ruled that training on legally obtained books was “quintessentially transformative” and qualified as fair use 2.
The court found no fair-use cover for downloading and keeping pirated books from pirate sites such as Library Genesis (LibGen), an online repository of unauthorized books, and Pirate Library Mirror (PiLiMi), a mirror site for pirated books, to build a central library. Those piracy claims continued and became the basis for settlement 2, 3.
Settlement eligibility required a title to appear on the court-approved “Works List,” meaning it was among the LibGen and PiLiMi files Anthropic downloaded, plus timely registration with the U.S. Copyright Office. That left about 482,460 eligible books out of more than 7 million copies downloaded 4.

The settlement could widen the gap between large and small AI firms

The US$1.5 billion payout sits below 1% of Anthropic’s US$183 billion valuation, so some observers see it as a manageable business expense 5.
Smaller AI companies may struggle to fund similar “bet-the-company” litigation, which can strengthen the lead of well-funded rivals 6.
Some describe the case as a “warning shot” that may steer companies away from pirated data toward licensing deals with publishers and authors. That shift may reward established copyright holders and favor large AI firms that can license at scale 1, 6.