Anthropic’s $1.5 billion settlement with authors and publishers marks a huge financial reckoning for one AI company, but it does not settle the bigger fight over whether copyrighted books can legally be used to train AI systems.

A federal judge has now given final approval to the deal, allowing payments to begin in what has been described as the largest known copyright recovery in history.¹ The case grew out of claims that Anthropic used copyrighted books without permission while building its AI models.

The immediate outcome is clear: Anthropic will pay. Coverage of the ruling says authors are expected to receive about $3,000 per pirated book, with payments to go out "as promptly as possible" after attorneys take their share.² Anthropic has also signaled it wants the dispute finished, saying more than 91% of eligible authors and publishers have claimed a share of the settlement and that it is "looking forward to bringing this matter to a close."¹

But the legal meaning of the case is more complicated. According to reporting on the decision, the underlying court rulings drew a distinction between two separate issues: training AI on copyrighted text, and obtaining that text through piracy.³ Judge William Alsup previously found that training an AI model on copyrighted books counted as fair use, a major point in Anthropic’s favor.³ At the same time, he found Anthropic had illegally downloaded and stored millions of books from pirate sources, exposing the company to potentially steep damages.³

That split helps explain why reactions are mixed. Supporters of the settlement can point to a historic payout and tangible compensation for writers.¹ Critics, including some authors in other related disputes, argue that roughly $3,000 per book is too little and that the broader copyright question remains unresolved.¹ Because the case settled, the fair-use ruling will not be tested on appeal, leaving other lawsuits against companies including Google, Meta, and OpenAI to shape the next phase of the battle.³