Anthropic Resolves Legal Case Over AI-Generated Book Piracy

Anthropic has reached a resolution in a significant copyright dispute with US authors, agreeing to a proposed class action settlement that avoids a potentially costly trial. The agreement, filed in court documents this Tuesday, stems from allegations that the AI company trained its Claude models using pirated literary works.

The settlement details remain confidential, though the case originated from claims by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson. Their 2023 lawsuit alleged Anthropic utilized an open-source dataset containing unauthorized copies of copyrighted materials for AI model training.

While Anthropic previously secured an important ruling in June - with Judge William Alsup determining that training AI on legally acquired books constitutes fair use - the decision preserved the possibility of additional legal challenges. This July, the judge certified a class action alleging Anthropic engaged in "Napster-style downloading" of copyrighted works.

The impending trial, scheduled for December, could have exposed Anthropic to extraordinary financial risks, with potential damages reportedly ranging into the trillions. According to court documents, the settlement agreement should receive final approval by September 3.

"This groundbreaking resolution will provide meaningful benefits for all affected authors," stated Justin Nelson, legal representative for the plaintiffs. He indicated that specific settlement terms will be disclosed in the coming weeks. Anthropic representatives declined to offer any public comment regarding the settlement agreement.