The class action lawsuit, filed Friday night in a San Francisco federal court, says Nvidia used the copyrighted books to train its new NeMo platform, created by the chipmaker as a «fast and affordable way to implement generative AI».
Writers Brian Keane, Adi Nazemian, and Stuart O’Nan they say, that their work was part of a dataset of approximately 196,640 books, and helped train NeMo to simulate normal written language before being removed in October by «due to alleged copyright infringement by». They note that such actions indicate that Nvidia «has recognized»’s infringement.
The authors are seeking unspecified damages for people in the United States whose copyrighted works helped train NeMo’s large speech models over the past three years. Among the works covered by the lawsuit are — Keane’s 2008 novel «Ghost Walk», Nazemian’s 2019 novel «Like a Love Story», and O’Nan’s 2007 short story «Last Night at the Lobster».
Earlier, a similar lawsuit was filed by OpenAI — from the creator of «Game of Thrones» George Martin and other authors, who accused the ChatGPT developer of copying their work «collectively, without permission or consideration».
Nvidia, meanwhile, is benefiting from the relentless interest of investors in AI — since the end of 2022, the chipmaker’s shares have increased by 600%. At the moment, the company’s market value is close to $2.2 trillion.