A federal judge in San Francisco gave final approval on Monday to Anthropic’s $1.5 billion copyright settlement with a class of authors who sued the company for using their books without permission to train its Claude AI models. U.S.
District Judge Araceli Martinez-Olguin signed off on the deal, formally closing the Bartz v. Anthropic class action and cementing it as the largest known copyright settlement in United States history.
The ruling rejected arguments from objectors who claimed the payout was insufficient, with the judge determining that the settlement terms were fair, reasonable, and adequate given the uncertainties of further litigation.
The case centred on Anthropic’s use of two pirated book datasets — LibGen and PiLiMi — to train Claude. The settlement agreement acknowledges that approximately 500,000 titles out of the roughly 7 million books Anthropic reportedly downloaded from those sources are covered under the class action.
Each class member is estimated to receive approximately $3,100 per qualifying work. Beyond the financial terms, Anthropic is required to destroy the two pirated datasets and all derivative copies within 30 days of final judgment — a structural remedy that goes further than a cash payment alone and addresses the original harm at its root.
For a clear breakdown of what the settlement means for authors and how the payout structure works, this overview is worth watching: https://www.youtube.com/watch?v=LPs-SNuOqho
The Payment Timeline and Who Gets What
The $1.5 billion total is being distributed across a structured schedule rather than as a single lump sum. Anthropic paid an initial $300 million tranche within five days of final approval. A further $450 million instalment is due by September 25, 2026, with two additional $450 million payments due on the first and second anniversaries of the preliminary approval order.
Interest accrues on the unpaid balance, giving class members additional compensation for the time elapsed since the litigation began. The claim deadline passed on March 30, 2026, and the Authors Guild reported that approximately 91.3 percent of the books covered by the settlement had been claimed — a notably high participation rate for a copyright class action of this scale.
Not every author who was eligible chose to participate. An additional 350 class members opted out of the settlement, and 53 filed formal objections. Those who opted out retain the right to bring their own independent claims against Anthropic, and several have already done so — a group of opt-out authors filed a separate copyright suit in May 2026 requesting a jury trial.
What This Means Beyond Anthropic
The Bartz v. Anthropic settlement is being watched closely across the AI industry because it is the first major US copyright case involving AI training data to reach a full resolution. It does not, however, resolve the broader legal question of whether using copyrighted works to train AI models constitutes fair use — that question remains live in several other active cases.
The New York Times’ lawsuit against OpenAI, which escalated sharply in July 2026 with new allegations that OpenAI concealed evidence of tools capable of identifying copyrighted journalism in ChatGPT outputs, remains the most closely watched of the outstanding cases.
OpenAI has denied those claims, but the Bartz settlement now provides plaintiffs in other AI copyright cases with a concrete financial precedent — and a very large number — to anchor future settlement negotiations around.
For Anthropic, final approval closes the most significant legal liability the company was carrying and removes the uncertainty that a $1.5 billion contingent obligation creates for a company that recently closed a $3.5 billion fundraising round.
The settlement cost is substantial, but the alternative — years of continued litigation with no guaranteed outcome — carried risks that the company’s board and investors clearly decided were greater than the cost of resolution.
Quick Links:
