Particle.news

Download on the App Store

Anthropic Seeks Approval of $1.5 Billion Settlement With Authors Over Pirated Books

The filing follows a June ruling that distinguished lawful AI training from unlawful mass downloading of pirated books.

Overview

  • Anthropic and the plaintiffs asked a San Francisco federal judge to grant preliminary approval, with a hearing set for Sept. 8.
  • The deal would pay roughly $3,000 per covered work for about 500,000 works, with additional payments if the list grows, and authors can opt out.
  • Anthropic agreed to destroy downloaded copies obtained from shadow libraries such as LibGen, PiLiMi and Books3, and to fund the settlement in four installments starting with $300 million after preliminary approval.
  • The agreement includes no admission of liability and resolves the authors’ legacy acquisition claims while leaving potential claims over model outputs intact.
  • Judge William Alsup previously held that training can qualify as fair use but faulted Anthropic for stockpiling over seven million pirated books, a posture that raised trial exposure that plaintiffs’ counsel now tout as yielding the largest publicly reported copyright recovery.