CALIFORNIA (WOWO) — A federal judge in San Francisco has granted final approval to a record-breaking $1.5 billion settlement resolving a landmark class-action copyright lawsuit against artificial intelligence developer Anthropic.

The agreement marks the largest known copyright recovery in legal history. Under the terms of the settlement, the San Francisco-based maker of the Claude AI chatbot will pay approximately $3,000 for each book that was illegally obtained and ingested into its development system. Over 91% of the more than 482,000 registered titles covered by the case have already been claimed by affected authors and publishers.

The lawsuit, originally filed in 2024 by bestselling authors, alleged that Anthropic unlawfully downloaded millions of copyrighted works from illicit online “shadow libraries” to train its generative language models.

While federal courts ruled that training AI systems on text constitutes “fair use” under U.S. copyright law, the judge found the company liable for illegally acquiring and storing pirated digital files.

Under the court-approved deal, Anthropic is required to completely destroy all illicitly obtained book files in its possession. Legal experts note that while the $1.5 billion payout creates a high price tag for past piracy, the ruling confirms that AI companies can continue using text for training models provided the data is sourced legally.