Nvidia Lawsuit: GeForce Maker Accused of Copyright Infringement By Authors of AI Models

Nvidia Corporation, a software and technology company, is facing a lawsuit after three authors reportedly accused it of copyright infringement. It allegedly used the authors' content without permission to train its NeMo, an artificial intelligence (AI) platform.

According to Fox Business, Stewart O'Nan, Brian Keene, and Abdi Nazemian were the three authors who sued Nvidia Corporation. In their suit, they claimed their works were added to a dataset of 196,640 books that were utilized to train NeMo to be a mock-up of ordinary written language before these content were withdrawn in October because of "reported copyright infringement."

Lawsuit Filing at the Federal Court

O'Nan, Keene, and Nazemian were said to have proposed a class action lawsuit against Nvidia, which was filed on the evening of March 8 at a federal court in San Francisco. They told the Santa Clara, California-headquartered software firm "admitted" it trained NeMo using their dataset, thus infringing their copyrights.

It was added that this lawsuit is similar to other cases that were filed by others for AI copyright infringement. At any rate, this suit is seeking unspecified damages for American individuals whose copyrighted works were used to train NeMo's large language models (LLMs) in the past three years.

Alleged Copyrighted Content

Based on the lawsuit, some of the works that were put in the dataset are Nazemian's "Like a Love Story" novel that was published in 2019, Keene's "Ghost Walk" 2008 novel, and O'Nan's "Last Night at the Lobster" novella that was released in 2007.

The authors said these books were added to a data set called "The Pile," where a collection of books known as "Books3" was stored. They further claimed Nvidia already admitted to having trained its NeMo Megatron AI models on Books3 and The Pile.

Finally, PYMNTS reported that the plaintiffs argued that the mere fact of the books being removed is already tantamount to an admission by Nvidia that it indeed trained NeMo on the dataset, so it has infringed their copyrights. The publication contacted Nvidia to ask about the lawsuit, but it reportedly declined to comment.

Photo by: Coolcaesar/Wikimedia Commons(CC BY-SA 4.0)