- Bloomsbury is among beneficiaries of a $1.5 billion AI copyright settlement involving Anthropic
- The settlement resolves claims Anthropic used copyrighted books without permission for AI training
- Bloomsbury and affected authors will receive about $19 million after legal fees deduction
Britain's Bloomsbury, the publishing house behind the Harry Potter books by JK Rowling this week confirmed that it was among the beneficiaries of a landmark $1.5 billion settlement. The settlement resolves claims that Anthropic, the artificial intelligence company behind Claude, used copyrighted books to train its AI models without purchasing the content. Other than Rowling, Bloomsbury is also the publisher for bestselling novelists Sarah J. Maas and Susanna Clarke.
The British publishing company along with the group of affected authors is expected to receive about $19m in settlement, post a 10% deduction for attorney fees and other expenses.
A US court identified 14,087 of its titles covered by the settlement, with proposed compensation of about $3,000 per title, split equally between the author and publisher.
The settlement stems from a class-action lawsuit, approved by a federal judge this Monday. It was brought forth by authors who alleged Anthropic used their books without permission to train its Claude AI chatbot. The settlement is the largest known copyright payout in US history. Bestselling thriller writer Andrea Bartz first brought the lawsuit with two other authors in 2024.
The London-based company expects to receive the cash from the settlement in instalments, potentially starting in the second half of this fiscal year, and will share half the proceeds with the authors.
The settlement does not require Anthropic to admit wrongdoing, but it marks one of the clearest financial consequences yet for AI companies accused of using copyrighted material without permission. A growing number of publishers, authors, news organisations and artists have filed lawsuits against AI companies. They allege that their work was copied to build powerful AI models without consent or compensation.
Anthropic has maintained that using publicly available material for AI training falls under the legal doctrine of fair use. The company, however, has also been signing licensing agreements with publishers and media companies as the legal scrutiny around AI training data intensifies.
This outcome is expected to be a shot in the arm for the publishing industry, where concerns over AI training have been mounting. While the payment offers compensation for past use, the broader legal battle over whether AI companies can freely train models on copyrighted works without licences is far from settled.
Several other copyright lawsuits involving AI developers, including cases against ChatGPT maker OpenAI and Meta, are still making their way through US courts and could further shape how copyrighted content is used in the AI era.
