The artificial-intelligence company Anthropic will pay $1.5 billion to a group of authors who sued the firm for training its large language models on pirated copies of their books, the New York Times reports.
In June, U.S. District Judge William Alsup ordered Anthropic to stand trial in a class-action copyright infringement suit brought by the authors, ruling that while using their books to train the models, named Claude, was fair use, the firm’s use of pirated books was unlawful.
Last month, the company and the authors agreed to settle the case. Justin Nelson, a lawyer representing the class of writers, said in a statement, “This historic settlement will benefit all class members. We look forward to announcing details of the settlement in the coming weeks.”
Under the settlement, Anthropic will pay 500,000 authors $3,000 per book. The settlement must still be approved by Alsup.
Aparna Sridhar, deputy general counsel for Anthropic, said in a statement to the Times, “Today’s settlement, if approved, will resolve the plaintiffs’ remaining legacy claims. We remain committed to developing safe A.I. systems that help people and organizations extend their capabilities, advance scientific discovery and solve complex problems.”
Nelson, lawyer for the authors, said, “This settlement sends a powerful message to A.I. companies and creators alike that taking copyrighted works from these pirate websites is wrong.”
Michael Schaub is a contributing writer.
