Judge: Nvidia's Shadow Library Scripts 'Have No Other Purpose' Than Infringement
6 hours ago
- #Copyright Infringement
- #AI Training
- #NVIDIA
- NVIDIA's revenue has surged due to high demand for its AI chips and data center services, benefiting from the AI boom.
- Authors have sued NVIDIA for copyright infringement, alleging its AI models were trained on the pirated Books3 dataset and that NVIDIA had contacts with Anna's Archive.
- NVIDIA filed a motion to dismiss the lawsuit, calling the allegations speculative and legally insufficient, but Judge Jon Tigar denied most of the dismissal request.
- The contributory copyright infringement claim survives, with the court ruling that scripts distributed by NVIDIA to download The Pile dataset have no purpose other than enabling infringement.
- Direct copyright infringement claims related to BitTorrent were not dismissed, with the judge noting BitTorrent is merely a tool, not a library or dataset.
- The vicarious copyright infringement claim was dismissed, but authors have 21 days to refile, and the lawsuit against NVIDIA is moving forward.
- Similar AI training lawsuits are growing, including a new case against Meta and Mark Zuckerberg for training on pirated books.