Anthropic has arrived at an initial agreement in a class action suit initiated by a collective of notable authors, signifying a significant shift in one of the most important AI copyright disputes in history. This action will enable Anthropic to sidestep what might have been a financially catastrophic verdict in court.
The agreement is anticipated to be finalized on September 3, with additional information to follow, as per a legal document released Tuesday. Anthropic opted not to provide any comments. “This historic agreement will be advantageous for all class members. We are eager to share further details of the settlement in the coming weeks,” stated Justin Nelson, an attorney for the plaintiffs, in a declaration to WIRED.
In 2024, three authors, Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, took legal action against Anthropic, claiming that the startup unlawfully utilized their works to train its AI models. In June, California district court judge William Alsup delivered a summary judgment in Bartz v. Anthropic that mostly favored Anthropic, determining that the company’s use of the books was “fair use” and therefore permissible.
However, the judge ruled that the method by which Anthropic obtained some of the works, by downloading them via so-called shadow libraries, including a infamous site known as LibGen, amounted to piracy. Alsup decided that the authors could still pursue Anthropic in a class action for pirating their works; the legal confrontation was set to commence in December.
Statutory damages for such piracy begin at $750 for each violated work, under US copyright law. Given that the library of books collected by Anthropic was believed to include about 7 million works, the AI organization was potentially facing court-imposed fines reaching billions, perhaps exceeding $1 trillion.
“It’s a surprising change of circumstances, considering how Anthropic was vigorously contesting this case in two courts. Moreover, the company had recently recruited a new trial team,” remarked Edward Lee, a law professor at Santa Clara University who closely follows AI copyright affairs. “However, they had limited defenses at trial, based on Judge Alsup’s ruling. Thus, Anthropic was facing the looming threat of statutory damages in ‘doomsday’ sums.”
The majority of authors who might have been included in the class action were just beginning to receive notifications that they were eligible to participate. The Authors Guild, a professional writers’ trade organization, disseminated a notice earlier this month informing authors of their possible eligibility, while lawyers for the plaintiffs were expected to submit a “list of affected works” to the court on September 1. This implies that many of these writers were not involved in the negotiations that occurred.
“The primary question is whether there will be a significant backlash from within the author class once the settlement terms are made public,” noted James Grimmelmann, a professor of digital and internet law at Cornell University. “That will serve as an important gauge of the sentiment among copyright owners.”
Anthropic continues to confront a series of other copyright-related legal issues. One of the most notable disputes involves several major record labels, including Universal Music Group, which allege that the company unlawfully trained its AI programs on copyrighted lyrics. The plaintiffs have recently filed to revise their case to assert that Anthropic used the peer-to-peer file sharing service BitTorrent to illegally download songs.
Settlements do not establish legal precedents, but the specifics of this case are likely to be closely monitored as numerous other prominent AI copyright cases progress through the judicial system.
Update: 8/26/25, 11:40 pm EST: This article has been updated to include comments from an attorney representing the plaintiffs.
