Anthropic Settles AI Book Training Lawsuit | Authors Guild

Anthropic Reaches Settlement in Authors' Copyright Lawsuit

A class action lawsuit brought by a collective of both fiction and nonfiction authors against Anthropic has been resolved, as evidenced by a court filing submitted to the Ninth Circuit Court of Appeals on Tuesday.

Previously, Anthropic had achieved a limited success with a ruling in a lower court. The company was actively pursuing an appeal of that initial decision.

The specifics of the settlement remain confidential, and a response from Anthropic to inquiries for comment was not immediately available.

The Core of the Dispute: Training Data and Fair Use

The lawsuit, formally known as Bartz v. Anthropic, centers on the utilization of copyrighted books as source material for training Anthropic’s large language models.

While the court had determined that Anthropic’s application of the books constituted fair use, the presence of pirated copies within the dataset resulted in potential financial repercussions for the company.

Anthropic's Initial Response to the Court Ruling

Despite the complexities, Anthropic had initially celebrated the earlier court decision, characterizing it as a positive outcome for the field of generative AI.

Following the June ruling, the company communicated to NPR, stating: “Our acquisition of books was solely for the purpose of developing large language models, and the court affirmed that this use falls under fair use principles.”

Plaintiffs' Counsel Expresses Satisfaction

In a statement provided to TechCrunch, legal representatives for the plaintiffs expressed their approval of the settlement.

Attorney Justin Nelson commented, “This landmark settlement will provide benefits to all members of the class.” He further indicated that further details regarding the settlement would be disclosed in the near future.

Note: This article was updated on August 27th at 8:37am ET to include a statement from the plaintiffs’ legal team.