A trio of authors and journalists led a class-action lawsuit against AI start-up Anthropic for allegedly using hundreds of thousands of copyrighted books to train its Claude chatbots.

The lawsuit filed on 19 August in a California court stated the AI company “intentionally downloaded known pirated copies of books from the internet.

It also stated Anthropic made unlicensed copies of the books, which were used to “digest and analyse the copyrighted expression”.

“The end result is a model built on the work of thousands of authors, meant to mimic the syntax, style, and themes of the copyrighted works on which it was trained.”

While the Google- and Amazon-backed start-up has touted its safe and responsible use of AI, the authors stated it had “built a multibillion-dollar business by stealing hundreds of thousands of copyrighted books”.

Claimants say Anthropic pirated books by accessing a dataset which included illegal copies of books instead of “obtaining permission and paying a fair price for the creations it exploits”, Anthropic pirated them”.

The lawsuit is led by Andrea Bartz, Charles Graeber and Kirk Wallace Johnson, but includes thousands of unnamed authors. They are seeking to stop Anthropic from using their works and an unspecified amount of monetary damages.

In May, a group of eight US newspapers filed a lawsuit against Microsoft and OpenAI claiming the companies used their articles to train generative AI models without permission.