Nvidia is in hot water with a big class-action lawsuit claiming it used pirated books to train its AI models. The lawsuit alleges that the company actively attempted to acquire millions of copyrighted books and academic papers.
This was done from sites such as Anna’s Archive, LibGen, and Sci-Hub. Internal emails reportedly show Nvidia looking into these sources to enhance its AI capabilities. This has sparked questions about copyright rules and the ethics of the tech industry.
Complaint filed against Nvidia for using pirated books
The updated lawsuit in U.S. District Court claims Nvidia’s top executives greenlit reaching out to Anna’s Archive, a site famous for giving free access to millions of copyrighted books and papers. Emails in the complaint say someone from Nvidia’s data strategy team contacted the archive to see if its library could be used to train large language models.
The suit alleges that “competitive pressures” in the AI world pushed Nvidia to go after pirated datasets. Besides Anna’s Archive, Nvidia reportedly tapped other sources like LibGen, Sci-Hub, and Z-Library to grab copyrighted material. Although Anna’s Archive allegedly warned Nvidia about legal risks, they moved ahead, gaining access to around 500 terabytes of data.
The lawsuit also says some of the material Nvidia accessed came through the Internet Archive’s controlled digital lending system, which is already tied up in copyright battles. On top of that, the filing claims Nvidia shared scripts and tools with corporate clients to automatically download datasets packed with pirated books, spreading the alleged copyright violations beyond just internal use.
Nvidia, however, pushed back, saying its AI model NeMo was built under fair use and followed copyright rules. They’re not the only ones under the microscope. Spotify confirmed in December 2025 that it’s looking into claims involving 300 terabytes of scraped data from Anna’s Archive.
Originally reported by Rishabh Shandilya on Mandatory.
