A federal court in San Francisco has approved a record-breaking copyright litigation settlement agreement, with AI giant Anthropic agreeing to pay up to $1.5 billion in compensation to book authors. This class-action lawsuit stemmed from the company's large-scale download of unauthorized book works from pirated databases for AI model training in recent years.

According to the details of the court-approved agreement, nearly 480,000 works have been successfully claimed by the authors, who are entitled to approximately $3,000 per book in compensation. In addition to paying this historically highest copyright settlement, Anthropic is also required to completely destroy all pirated files.

Piracy Is Penalized, but AI Training Is Protected

Although the compensation amount is significant, the legal community generally believes this ruling actually represents a key legal victory for the AI industry. The presiding judge previously clearly stated that using legally obtained books to train AI constitutes fair use, and its innovativeness and transformative significance are evident.

This means that the compensation targets the pirated behavior of illegally acquiring resources, not the AI training itself. Although authors retain the right to pursue legal action against AI-generated infringing content and the company's future actions, the clarification of the fair use principle has cleared away a major legal obstacle for the entire AI field.