Anthropic, a startup in artificial intelligence backed by Google (GOOGL.US) and Amazon (AMZN.US), is facing a lawsuit over data copyright issues.

Intelligensia reported that Anthropic, a startup artificial intelligence company, is facing a class action lawsuit in the federal court in California, with three writers suing the San Francisco-based company for allegedly using their work and that of tens of thousands of others to train its artificial intelligence chatbot, Claude.

On Monday, the writers and journalists Andrea Bartz, Charles Graeber and Kirk Wallace Johnson filed the suit, alleging that “Anthropic has built a multibillion-dollar business by stealing tens of thousands of copyrighted books without permission or payment, and in violation of copyright law.”

The plaintiffs claim that their work and that of others was used to train the company’s large language model, GPT-3, which powers its flagship product, Claude. The plaintiffs claim: “A key part of Anthropic’s business model — and its flagship product, the ‘Claude’ series of large language models — is the wholesale theft of copyrighted works.”

The lawsuit is the second against the company, which is backed by Amazon (AMZN.US) and Alphabet’s (GOOGL.US) Google. The UK and US antitrust regulators are currently investigating the company’s partnerships with Google and Amazon.

The plaintiffs claim that although Anthropic has kept the sources of its training data secret, they add that one of the creators of The Pile, a data set called Books3, is Shawn Presser, an independent data provider who created a dataset of heavily pirated books.

The plaintiffs are seeking an undisclosed amount of money in damages and are seeking a court order to permanently enjoin Anthropic from engaging in the alleged infringement.

The lawsuit is one of several legal actions against companies developing artificial intelligence models that are alleged to have used copyrighted materials. Different groups of authors have sued OpenAI, backed by Microsoft (MSFT.US), and Meta (META.US), alleging that they have used work to train large language models that power their chatbots.