US judge approves Anthropic’s $1.5bn AI copyright settlement

Anthropic is set to pay $1.5 billion to resolve a class-action copyright case brought by authors over the training of its Claude artificial intelligence chatbot, Reuters reported Monday. U.S. District Judge Araceli Martinez-Olguin approved the settlement, which Reuters said is believed to be the largest copyright award of its kind in the United States.

The case centered on allegations that Anthropic used authors’ works without proper authorization as part of the data used to train Claude. The settlement makes Anthropic the first major U.S. defendant to resolve one of the current wave of lawsuits testing how copyright law applies to generative AI training, according to Reuters.

The dispute forms part of a broader legal fight between rights holders and technology companies building AI models. Authors, publishers and news organizations have sued AI developers, arguing that copyrighted books, articles and other material were incorporated into training datasets without permission or adequate compensation.

Training an AI model generally involves feeding large volumes of text and other data into software systems so they can identify statistical patterns and generate new outputs in response to user prompts. Rights holders have argued that the process can rely on protected works at industrial scale. AI companies have often defended training practices by invoking fair use, a doctrine that permits some uses of copyrighted material depending on factors such as purpose, market effect and the nature of the work.

Anthropic Deputy General Counsel Aparna Sridhar said in a statement cited by Reuters that the settlement was reached in 2025 after a court ruling that training AI on books qualifies as fair use under copyright law. She said that ruling “remains the law today.” Sridhar also said more than 91% of authors and publishers covered by the settlement had claimed a share of the payment.

Justin Nelson, lead counsel for the plaintiffs, described the agreement as a “historic settlement,” Reuters reported. “It is the largest known copyright recovery in history,” Nelson said, adding that the plaintiffs’ side expected to distribute money to the class as promptly as possible.

The settlement does not end the wider legal uncertainty around AI and copyright. U.S. courts have reached differing views in related cases, Reuters and PYMNTS reported. U.S. District Judge William Alsup has said copyright law is intended to advance original authorship rather than shield authors from competition. U.S. District Judge Vince Chhabria has taken a more cautious view, warning that broad AI training practices could weaken the economic incentives supporting human creative production.

Daryl Lim, H. Laddie Montague Jr. Chair in Law at Penn State Dickinson Law, told PYMNTS in December that only a small number of companies can train frontier AI models at scale because they control the necessary compute, data, cloud infrastructure and distribution. “When you train frontier models, you need to ingest vast repositories of works that may include copyrighted works,” Lim said.

The approval gives Anthropic a path to close one high-profile legal exposure while other AI copyright cases continue through U.S. courts. For AI developers, publishers and investors, the settlement adds a concrete dollar figure to a legal risk that has become central to the economics of model training.

This story draws on original reporting from PYMNTS.