Particle.news
Download on the App Store

Major Publishers Sue Google Over Alleged Use of Millions of Books to Train Gemini

Plaintiffs say internal Google documents show the company knew of huge legal risks after copying publisher‑provided and scraped works to build its AI models, exposing Google to statutory damages and a court injunction.

Overview

  • A putative class action filed by Hachette, Cengage, Elsevier and author Scott Turow accuses Google of willfully copying millions of books and journal articles to train its Gemini models and seeks an injunction and statutory damages.
  • The complaint, filed in the Southern District of New York in mid‑July, alleges Google used works provided for Google Books and Play plus web‑scraped material from paywalled and piracy sites without publisher permission.
  • Plaintiffs cite internal Google documents that warned using publisher content for AI was “highly problematic” and referenced potential fines of “$10Bs‑$100Bs,” which they say shows the company knew it lacked authorization.
  • The suit says Gemini can produce verbatim and near‑verbatim passages, replacement chapters and close substitutes that displace sales and weaken the market for licensed books and academic articles.
  • The case adds to a wave of suits testing whether AI training on copyrighted works is lawful, and its outcome could force broader licensing, raise training costs for AI developers, and reshape how models are built.