On July 20, 2026, a US judge gave final approval to a $1.5 billion settlement from Anthropic, an artificial intelligence company. The company was sued for copying hundreds of thousands of books without asking and using them to train its AI — the same kind of technology behind chatbots like Claude.

The lawsuit started on August 19, 2024, when three authors — Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson — filed a complaint in a California federal court. According to the Authors Guild, Anthropic downloaded about 7 million copies of books to train its AI models. A judge first gave tentative approval to the settlement on September 25, 2025, and the final approval came from Judge Araceli Martínez-Olguín, who said the settlement provided "meaningful relief" to the affected authors and publishers.

Think of it like this: if someone photocopied your book thousands of times without telling you and used those copies to build a product they sell, you would probably want to be paid. That is essentially what this case was about.

The settlement covers about 482,000 copyrighted works. Around 91% of the authors and publishers affected have claimed their share of the money. Bloomsbury Publishing, which had 14,087 books listed in the settlement, said it expects to receive about $19 million (£14 million), after deductions.

The authors' lead lawyer, Justin Nelson, called it "the largest known copyright recovery in history." The lawyers' fees exceed $100 million, though a judge had previously reduced the amount, calling the original request too high. Nearly 120,000 authors and other copyright holders sought a share of the $1.5 billion fund.

For Bloomsbury, the payout works out to about $3,000 per book. After a roughly 10% deduction for legal fees and expenses, the publisher said it expects to receive the money in installments, possibly starting in the second half of this fiscal year. Bloomsbury confirmed the court's approval through an announcement on the London Stock Exchange under its ticker symbol, BMY. Anthropic's deputy general counsel, Aparna Sridhar, said the company was satisfied that more than 91% of those covered by the settlement have claimed their share.

This is the first major resolution among dozens of lawsuits filed in the US against AI companies for using copyrighted material without permission. Other groups involved in the case included the Textbook and Academic Authors Association and the Society of Authors, which supported the settlement.

The broader picture is that this settlement gives a real-dollar number — about $3,000 per book — that could shape how courts and lawmakers think about AI companies using copyrighted work. It also points to a wider shift: Bloomsbury, for example, announced a separate deal in 2025 to sell its academic works to AI companies for training, with authors choosing to participate and receiving royalties. So publishers are pursuing two paths at once — getting paid for past unauthorized use, and setting up new deals for the future. How that plays out will matter well beyond this one case.