The settlement is the largest known copyright payout in U.S. history.
The dispute reflects the intensifying legal battle over how generative AI models are trained. AI developers, including Anthropic, have argued that using publicly available copyrighted material qualifies as fair use under U.S. copyright law. Authors, publishers and other creators, however, maintain that AI companies should obtain licences or compensate rights holders before using protected works.
London:Bloomsbury Publishing will receive approximately £14 million ($19 million) to be shared with authors after a landmark $1.5 billion copyright settlement between AI startup Anthropic and hundreds of thousands of authors over the use of copyrighted works to train artificial intelligence models.
The London-based publisher said 14,087 titles from its catalogue were included in the settlement, with compensation averaging about $3,000 per title.
Payments are expected to be made in instalments, with the first tranche likely to be received during the second half of the company’s current financial year.
Following deductions of around 10% for legal fees and related expenses, Bloomsbury and participating authors are expected to receive a combined $19 million (£14 million).
The settlement resolves claims that Anthropic used copyrighted books without authorization to train its Claude family of AI chatbots. The lawsuit, filed in 2024 by novelist Andrea Bartz and two other authors, is the first major AI copyright case in the United States to reach a settlement.
U.S. District Judge Araceli Martínez-Olguín approved the agreement, describing it as providing “meaningful relief” to authors and publishers whose works were used. Approximately 91% of the 482,000 works covered under the settlement have already been claimed.
Calling it “the largest known copyright recovery in history,” lead plaintiffs’ attorney Justin Nelson said the agreement sets an important precedent as courts continue to hear dozens of lawsuits over AI companies’ use of copyrighted content.
The dispute reflects the intensifying legal battle over how generative AI models are trained. AI developers, including Anthropic, have argued that using publicly available copyrighted material qualifies as fair use under U.S. copyright law. Authors, publishers and other creators, however, maintain that AI companies should obtain licences or compensate rights holders before using protected works.
Anthropic welcomed the resolution. Deputy General Counsel Aparna Sridhar said the company was pleased that more than 91% of eligible authors and publishers had claimed their share of the settlement and looked forward to concluding the matter.
The development comes as Bloomsbury expands its own AI licensing strategy.
The publisher launched an AI licensing programme last year, allowing academic content to be licensed for training generative AI systems. Authors can voluntarily opt into the scheme and receive royalty payments, with Bloomsbury now exploring the inclusion of additional subject areas for AI training licences.