Final approval came through for a $1.5B class action settlement between AI company Anthropic and authors who said their copyrighted books were used to train the Claude model without permission. The sum is the largest ever paid to resolve an AI training data lawsuit.
The litigation, launched in 2024, covered approximately 500,000 books. Each author in the class will receive roughly $3,000 from the distribution plan approved by the court. The complaint accused Anthropic of systematically scraping copyrighted texts to assemble its training dataset without obtaining licenses.
The company did not admit any wrongdoing. It had argued that training AI on publicly available material falls under fair use, the same defense cited by OpenAI, Google, Meta, and others. Because the case was settled without a trial verdict, it creates no binding legal precedent.
Preliminary approval arrived in September 2025, with authors having until March 2026 to file claims. The final distribution plan adds interest payments on top of the $1.5B base fund.
Dozens of parallel cases remain active against other AI developers. OpenAI is defending multiple class actions from authors and publishers, and similar lawsuits targeting Meta and Google are also working through the courts. Since Anthropic settled rather than litigated to a judgment, the fundamental fair use question around AI training data remains legally untested at trial.
Anthropic has raised more than $14B and recently crossed a $5B annualized revenue run rate. The $1.5B settlement equals roughly 30% of that annual revenue figure.
