Anthropic $1.5bn author settlement caps AI copyright liability
A $1.5 billion court-approved settlement resolves Anthropic's flagship copyright case, shielding the Amazon-backed AI firm from potentially catastrophic trial damages while clarifying the cost of data retention missteps.
A U.S. federal judge has finalized a $1.5 billion settlement between Anthropic and a class of authors, resolving the largest known copyright lawsuit in U.S. history. The ruling closes a dispute over the artificial intelligence startup’s use of copyrighted books to train its Claude chatbot. Anthropic, which is backed by Amazon and Alphabet, continues to deny any wrongdoing.
For Anthropic and its investors, the settlement eliminates a massive balance sheet threat. The company faced a trial that could have yielded damages in the hundreds of billions of dollars after a previous court ruling exposed a critical flaw in its data practices.
While retired U.S. District Judge William Alsup previously ruled that training AI models on copyrighted texts constitutes fair use, he found Anthropic violated copyright law by stockpiling more than 7 million pirated books in a central digital library. Retaining those files beyond what was necessary for model training left the company legally exposed.
U.S. District Judge Araceli Martinez-Olguin rejected objections from some authors who argued the payout was too low, noting the settlement reasonably accounted for the severe risks of a trial. The court also awarded $101 million in attorney fees, significantly less than the $187.5 million requested. More than 91% of eligible authors and publishers have already claimed their share of the funds.
Anthropic framed the resolution as a pragmatic move following a legal victory on the underlying technology. “We reached this settlement in 2025 after the court’s landmark ruling that training AI on books is fair use under copyright law, which remains the law today,” Anthropic Deputy General Counsel Aparna Sridhar said.
The resolution of this litigation arrives as Anthropic navigates separate regulatory hurdles in the U.S. In June 2026, the government partially lifted a two-week suspension on Anthropic’s Claude Mythos 5 model. Access was restored to over 100 trusted organizations, including Fortune 500 companies and critical infrastructure operators, though the system remains off-limits to the general public.
Lead attorney for the authors, Justin Nelson, characterized the deal as the largest copyright recovery on record. For the broader AI sector, the Anthropic case establishes a clear financial benchmark for the consequences of improperly hoarding training data, even when the act of training itself is protected.
