Judge Approves Anthropic's $1.5B Copyright Settlement, Largest in U.S. History

A federal judge on Monday gave final approval to Anthropic's $1.5 billion settlement of a class action copyright lawsuit brought by authors and book publishers, concluding what is believed to be the largest copyright settlement in U.S. legal history¹.

Judge Araceli Martinez-Olguin of the U.S. District Court for the Northern District of California signed off on the deal, taking over after Judge William Alsup, who had issued preliminary approval last year, retired. The payout will deliver $3,000 per work across an estimated 500,000 works, shared among the authors and publishers who hold rights to them.

The settlement resolves a case rooted in how Anthropic assembled its training data. Alsup had ruled that Anthropic illegally downloaded and stored millions of copyrighted books. Specifically, the court found that Anthropic built its training library from two sources: books it purchased and scanned — which the court deemed acceptable — and books it downloaded from pirate sites including Library Genesis and Pirate Library Mirror. Alsup found the second method illegal on its own terms.

Critically, however, Alsup sided with Anthropic on the core legal question at the heart of the AI industry's copyright debate: he ruled that training an AI model on copyrighted text counts as fair use. That ruling has been widely seen as a turning point for the AI industry, according to TechCrunch.

Despite the record-setting payout, many authors and creators do not view the outcome as a win. The fair use finding on training means the $1.5 billion addresses the piracy channel through which Anthropic obtained books, not the act of using copyrighted material to build AI models.

ANALYSIS The bifurcated outcome — fair use for training, liability for piracy — creates a specific legal template: AI companies that acquire copyrighted training data through legitimate channels face no copyright liability under this ruling, while those that source material from unauthorized repositories remain exposed. The distinction between method of acquisition and method of use is the operative line the court drew.

The $3,000-per-work figure, spread across 500,000 works, may set expectations for future copyright negotiations between AI labs and rights holders, though the settlement's terms are specific to the facts of this case.

For Anthropic's competitors — including OpenAI, Meta, and Google, all of which face their own copyright litigation — the fair use finding on training is the more consequential element of this case. The financial penalty attaches to the piracy conduct, not to the model-training practice itself.