Photo by Josh Withers on Pexels
AI developer Anthropic has tentatively agreed to a $1.5 billion settlement, plus interest, in a class-action copyright lawsuit, potentially marking the largest payout in U.S. copyright litigation history. The settlement addresses allegations that Anthropic used copyrighted works to train its AI models without proper authorization. Plaintiff lawyers hailed the agreement as a “remarkable result.”
The proposed settlement, pending court approval on September 8th, could provide individual copyright holders with approximately $3,000 per work. This amount may fluctuate based on the volume of claims submitted, with the potential for Anthropic to increase the overall settlement if claims surpass 500,000 works.
Key terms of the settlement include Anthropic’s commitment to destroy any downloaded files and copies of the allegedly infringing material. Crucially, the agreement pertains solely to past conduct and does not provide Anthropic with a license for future AI training or relinquish any claims arising after August 25, 2025.
The lawsuit stemmed from accusations that Anthropic built its business on unlawfully utilizing copyrighted books. Although a judge previously determined that training AI on legally acquired books was permissible, the upcoming trial focused on allegations of utilizing pirated materials. In July, a federal judge certified a class action allowing authors to represent all US writers whose work was allegedly part of libraries of pirated books downloaded by Anthropic.
Anthropic maintains its dedication to responsible AI development aimed at empowering individuals and addressing complex challenges. According to a statement reported by The Verge, the company’s deputy general counsel, Aparna Sridhar, highlighted a June ruling that purportedly supports Anthropic’s AI training practices as fair use under copyright law.
This settlement occurs against the backdrop of increasing copyright infringement lawsuits targeting AI companies. It also coincides with an increase in collaborations between AI firms and media entities willing to license data for AI training purposes in exchange for remuneration. Anthropic also faces additional legal action from Reddit and Universal Music, alleging misuse of data and copyright infringement, respectively.
Authors and rightsholders can visit AnthropicCopyrightSettlement.com for more details.