Anthropic (Again) Under Accusation: $75 Million Demanded for Piracy and Copyright Infringement

Anthropic returns to court once again for copyright violations in artificial intelligence. As reported by the New York Post, the registration was filed in a California court: over a hundred authors are seeking compensation exceeding $75 million, as the company reportedly used pirated copies of their works to train its AI models.

According to the complaint, filed on June 17, the San Francisco-based company allegedly employed over 500 illegally obtained books, including Get Good with Money by Tiffany Aliche and Like Water for Chocolate by Laura Esquivel. Among the authors involved are Nolan Bushnell, co-founder of Atari, screenwriter Zachary Sklar, and Newbery Medal winner Donna Barba Higuera.

The lawsuit claims that Anthropic has violated the copyright of at least one hundred authors. The plaintiffs are asking for compensation of $150,000 for each pirated work, bringing the total value of the claim to over $75 million.

At the heart of the accusations is also the method by which the company allegedly obtained the material. According to the plaintiffs, Anthropic used BitTorrent to access the sites Library Genesis and Pirate Library Mirror, from which it downloaded millions of unauthorized copies of books. The works would then have been compiled into a central library used as a basis for training the AI assistant Claude.

This new legal initiative comes after a previous class action concluded with a $1.5 billion settlement in September. The authors involved in the current proceeding chose not to join that agreement and to initiate an independent action.

Under the previous agreement, Anthropic had agreed to eliminate copies of the books that the authors claimed were downloaded illegally. However, the settlement left open the possibility of new claims related to any copyright infringements linked to the content generated by AI models. In that procedure, each author received about $3,000 for each work involved.

According to the new lawsuit, the claims are not limited to the unauthorized acquisition of the books. The plaintiffs assert that Anthropic also redistributed the protected works, an element that differentiates this legal action from the previous class action.

Overall, considering both the class action and individual lawsuits, the authors claim that Anthropic used approximately 7 million creative works without authorization to fuel the development of its AI systems. The company will now have to respond to these new accusations in the federal court of California.