Publishers Sue Google Over Gemini AI Training Data
The lawsuit claims Google brazenly bypassed its scope-limited book agreements to feed millions of copyrighted works into its multi-billion-dollar AI.
Summary
Hachette Book Group, other publishers and author Scott Turow have sued Google, accusing the tech giant of willful copyright infringement.
The plaintiffs allege Google illegally bypassed scope-limited agreements for services like Google Books and Google Play to train its Gemini AI.
The suit warns that Gemini threatens human writers' livelihoods by generating rapid, cheap substitutes like knockoffs and alternative novel versions.
Hachette Book Group, Cengage Learning, Elsevier and bestselling author Scott Turow have accused Google of willful copyright infringement in a new lawsuit filed in the US District Court for the Southern District of New York. The plaintiffs allege that the tech giant used its access to vast libraries of digitized books and scholarly articles to train its Gemini generative AI platform. These digital resources were originally provided under strict agreements for limited search and retail functions through services like Google Books and Google Play. The publishers argue that Google brazenly violated these scope-limited partnerships to build a competing commercial product.
According to the filing, Google illegally copied countless texts and downloaded unauthorized web scrapes of the internet that included known pirate sources. The plaintiffs state that Google intentionally removed or altered copyright data to hide the fact that its models were trained on protected materials. The group claims that Google knew it lacked authorization to harvest the works for artificial intelligence development. Internal documents cited in the lawsuit allegedly reveal that Google engineers flagged the use of publisher-provided books as highly problematic and warned of massive potential fines reaching up to $100 billion USD.
The lawsuit emphasizes the severe market threat posed by generative AI tools trained on copyrighted media. Court documents note that Gemini can generate a 100-page murder mystery set in a quiet seaside town in just 20 minutes for a few cents. This capability creates direct market substitutes in the form of alternative versions of famous novels, replacement chapters for academic textbooks and inferior knockoffs. Publishers state that the scale and speed at which Gemini operates present an unprecedented challenge to the livelihoods of human writers.
Legal battles in the rapidly expanding artificial intelligence sector continue to escalate with this latest filing. Authors and media companies regularly push back against tech firms scraping their intellectual property. A recent TechCrunch report notes that Anthropic was hit with a massive $1.5 billion USD penalty for pirating the works it trained on. As courts debate the limits of fair use in the age of generative technology, the current Google lawsuit seeks class action status to secure damages and force the destruction of all infringing copies currently in the company’s possession.
