Anthropic Settles $1.5B AI Copyright Case, Agrees to Delete Pirated Files
A significant milestone in the battle over AI and copyright has been reached. Anthropic, an AI company, has agreed to a $1.5 billion settlement with authors and publishers over copyright infringement. The settlement, provisionally approved by US District Judge William Alsup, covers around 500,000 books and will see authors and publishers each receive an average of about $3,000 per work.
The settlement, which follows a class-action lawsuit brought by authors including Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, resolves claims of infringement up to August 25, 2025. It does not, however, release claims related to AI outputs, either past or future. Judge Alsup, who described the settlement as 'fair', will make a final decision after notifying the affected authors.
Anthropic has agreed to delete all files originating from pirate sources like Library Genesis (LibGen) and Pirate Library Mirror (PiLiMi), as well as any copies derived from them. The court accused Anthropic of 'Napster-like' piracy, emphasizing that fair use does not apply to piracy, even if the later use is transformative. The decision confirms that AI companies are bound by copyright when obtaining data, but the boundary between legality and illegality for publicly accessible internet content remains unclear.
The settlement is a significant precedent, putting pressure on AI companies to consider licensing copyrighted works for AI training. Funds will be distributed per work, not per claimant, with a committee developing resolution proposals in case of multiple claims. The plaintiffs welcomed the settlement, stating it brings them 'one step closer to real accountability for Anthropic' and sends a clear signal to all AI companies about copyright holders' rights.
