Publishers Sue Google Over Unauthorized AI Training

Hachette, Cengage, Elsevier and novelist Scott Turow have joined a class action against Google, filed this week, accusing the company of feeding their copyrighted books into its Gemini AI models without permission. The group, which also includes the author organization S.C.R.I.B.E., alleges Google didn't just use the material but tried to cover its tracks, stripping or altering copyright information to hide where the training data came from.

Google is hardly the first AI company to end up in this position. Meta, OpenAI and Anthropic are all fighting similar suits from writers and publishers who say their work was scraped and fed into large language models without a license or a paycheck. Two California rulings have already sided with the tech companies, finding that AI training counts as fair use under copyright law written long before anyone imagined machines writing sentences.

Anthropic, meanwhile, got hit with a $1.5 billion settlement for pirating training material, the largest copyright payout in U.S. history, though plenty of eligible authors turned down the money to keep fighting in court instead. Google's case now heads to the same battleground.