Anthropic to pay $1.5B in first US AI copyright settlement, does not formally admit wrongdoing

Anthropic has agreed to pay at least $1.5 billion to settle a class action lawsuit brought by book authors alleging copyright infringement, marking the first AI copyright settlement in the United States. The company will pay approximately $3,000 for each copyrighted work it allegedly pirated from shadow libraries like LibGen while gathering training data for its AI models, setting a significant precedent for how AI companies must compensate creators for unauthorized use of their intellectual property.

What you should know: This landmark settlement establishes the first legal precedent requiring AI companies to compensate copyright holders for unauthorized use of their work.

• The $1.5 billion figure represents an estimated $3,000 per copyrighted work that Anthropic allegedly pirated from shadow libraries including LibGen, a notorious site that illegally distributes copyrighted books.
• While substantial, the settlement amount is far below the potential damages Anthropic could have faced at trial, with expert estimates ranging into the billions or even over $1 trillion.
• Anthropic is not admitting wrongdoing or liability as part of the agreement.

The legal backdrop: The case emerged from a broader wave of copyright litigation targeting tech companies over AI training data practices.

• Authors Andrea Bartz, Kirk Wallace Johnson, and Charles Graeber filed the original lawsuit in 2024 in the US District Court for the Northern District of California.
• In June, senior district judge William Alsup ruled that Anthropic’s AI training was protected under “fair use” doctrine, but allowed the case to proceed specifically regarding the company’s use of pirated books.
• Judge Alsup determined that “Anthropic downloaded over seven million pirated copies of books, paid nothing, and kept these pirated copies in its library even after deciding it would not use them to train its AI.”

Why this matters: The settlement sends a clear signal to the AI industry that using pirated content for training data carries significant financial consequences.

• “This landmark settlement far surpasses any other known copyright recovery. It is the first of its kind in the AI era,” says co-lead plaintiffs’ counsel Justin Nelson of Susman Godfrey LLP.
• The precedent could influence how regulators and creative industries approach future legal debates over generative AI and intellectual property rights.
• Since this was an “opt-out” class action, dissatisfied authors can still exclude themselves to file individual lawsuits.

What’s next: Anthropic continues to face additional copyright challenges beyond this settlement.

• The company is defending against a lawsuit from major record labels including Universal Music Group, which alleges unauthorized use of copyrighted lyrics to train its Claude chatbot.
• Plaintiffs in the music case are attempting to amend their complaint to include allegations that Anthropic used BitTorrent, a peer-to-peer file sharing service, to illegally download songs.
• The music industry lawyers have indicated they may file a separate piracy lawsuit if they’re not permitted to amend their current complaint.

What they’re saying: Both sides are framing the settlement as validation of their positions while avoiding admission of wrongdoing.

• “This settlement sends a powerful message to AI companies and creators alike that taking copyrighted works from these pirate websites is wrong,” Nelson stated.
• Anthropic deputy general counsel Aparna Sridhar emphasized the company’s continued commitment: “We remain committed to developing safe AI systems that help people and organizations extend their capabilities, advance scientific discovery, and solve complex problems.”