NVIDIA is said to have approved access to pirated books for AI training. The allegations come from an expanded class action lawsuit by several US authors.
• NVIDIA is said to have actively contacted Shadow Library
• Internal documents come from expanded class action lawsuit by US authors
• NVIDIA employee probably asked for 500 terabytes of book data
Class action lawsuit with explosive allegations
The chip company NVIDIA is facing serious allegations. According to an amended statement of claim filed in the United States District Court for the Northern District of California on January 17, 2026, the company is said to have actively contacted the so-called shadow library “Anna’s Archive”. The goal was to gain access to huge amounts of copyrighted books for training your own AI models.
As the magazine TorrentFreak, which specializes in copyright issues, reports, internal emails and documents show that a member of NVIDIA’s data strategy team contacted the operators of the piracy library. The employee inquired about the possibility of quick access to around 500 terabytes of data in order to integrate it into the training data for NVIDIA’s own Large Language Models.
Warning ignored – permission granted
Particularly explosive: Anna’s Archive is said to have expressly informed NVIDIA that the content offered was obtained illegally and asked whether there was internal authorization for access. According to the lawsuit, this approval was given by NVIDIA management within a week.
Anna’s Archive then offered the group access to about 500 terabytes of data, including millions of books that are normally only accessible through restricted digital lending systems such as the Internet Archive. The Shadow Library is said to have charged more than $10,000 for so-called express access. It is not clear from the documents whether NVIDIA actually paid.
Further allegations and background to the lawsuit
According to “heise online”, the original lawsuit was filed in early 2024 by several US authors, including Abdi Nazemian, Brian Keene and Stewart O’Nan. They accused NVIDIA of using their copyrighted works from the Books3 dataset without permission to train AI models such as NeMo, Megatron and Retro-48B. At the time, NVIDIA defended itself with the argument of the “fair use” rule – texts would only serve as statistical patterns for AI systems.
However, the now expanded statement of claim goes much further. In addition to Anna’s Archive, NVIDIA is also said to have downloaded and used content from LibGen, Sci-Hub and Z-Library. Additionally, the lawsuit accuses the company of distributing scripts and tools to business customers that allowed them to automatically download a dataset called “The Pile” – a dataset that includes public domain works as well as the pirated Books3 dataset. The plaintiffs see this as a case of indirect copyright infringement, as NVIDIA is said to have benefited indirectly from the distribution.
As TorrentFreak points out, this is the first time that email correspondence between a major US technology company and Anna’s Archive has been made public. The case joins a growing number of lawsuits against tech companies over the use of copyrighted content for AI training – similar proceedings are ongoing against OpenAI and Meta, among others.
D. Maier / editorial team finanzen.net
By the way: Meta Platforms (ex Facebook) and other US stocks can even be traded on finanzen.net ZERO until 11 p.m. (without order fees, plus spreads). Open a depot now for free and secure a new customer bonus!
Selected leverage products on Meta Platforms (ex Facebook)
With knock-outs, speculative investors can participate disproportionately in price movements. Simply select the lever you want and we will show you suitable open-end products on Meta Platforms (ex Facebook)
The leverage must be between 2 and 20
Advertising
