Publishers accuse Google of training AI with copyrighted works

Hachette Book Group, Cengage Learning, Elsevier, and writer Scott Turow, along with his publishing company S.C.R.I.B.E., have filed a lawsuit in New York against Google. They allege that Google secretly copied millions of works from its digital library, Google Books, and other services, which were provided for limited use, to train its AI model, Gemini. The plaintiffs claim that Gemini can create books and compete with human authors at an unprecedented scale and speed, and that it even imitates the stylistic elements and creative choices of specific authors. They argue that content generated by Gemini directly competes with the original authors' works. The lawsuit seeks a court order to stop these practices and unspecified damages. This legal action is part of a series of copyright infringement lawsuits against AI companies. Several of these publishers, including Hachette, Cengage, Elsevier, and Scott Turow, had previously sued Meta in May in a New York court for similar reasons. In a related case, Anthropic, developer of the Claude AI models, reached an agreement in September to pay at least $1.5 billion to authors and publishers who sued it for illegally downloading millions of books. However, a judge in that case ruled that feeding a generative AI model with works theoretically protected by copyright did not constitute an infringement.