IP Law Daily, COPYRIGHT NEWS—NVIDIA utilized ‘mass download-and-ingest‘ pipeline to download YouTube videos without authorization, a new complaint alleges, (Feb 3, 2026)

By Matthew Hersh, J.D

The company allegedly bypassed YouTube technological measures in order to build a training database for generative AI purposes.

A multi-billion dollar technology company and developer of artificial intelligence products allegedly bypassed YouTube’s anti-circumvention technical measures in order to download “80 years-worth of videos per day” to build a database of videos in violation of the Digital Millennium Copyright Act, a new federal lawsuit alleges. The lawsuit, which is brought by a relatively modestly-followed YouTube podcaster on behalf of a putative nationwide class of content creators who are similarly situated, alleges that the company built this dataset in order to train a massive generative artificial intelligence model (Youngblood v. Nvidia Corp., No. 5:26-cv-00916 (N.D. Cal. Jan. 29, 2026)).

The single named plaintiff in the lawsuit, Devin Youngblood, is an Alabama resident who owns and operates the YouTube channel “The Youngblood’s Podcast.” The podcast, which had just shy of 5,000 subscribers as of the date this article was written, consists of a series of weekly audio and video episodes in which Youngblood and his wife respond to listener-submitted questions and discuss “topics including marriage, parenting, family relationships, mental health, and personal experiences drawn from their own lives.” The content “is presented as candid; unscripted commentary based on the hosts’ personal perspectives and family experiences and is distributed via YouTube,” the complaint describes.

The lawsuit takes aim at NVIDIA, a publicly traded technology company with a present market capitalization of over $4 trillion and annual revenues over $100 billion. The company’s website describes its core markets as artificial intelligence, design and simulation, high-performance computing, gaming and creating, automotive, robotics and edge AI, and data center and cloud computing. Central to this lawsuit is a product called Cosmos, which NVIDIA describes a platform to accelerate the development of physical AI for autonomous vehicles, robots, and “video analytics AI agents.”

The complaint alleges that NVIDIA set out to aggregate “a huge curated video dataset” for the purposes of its generative modelling in connection with the Cosmos project. To do this, the complaint alleges, NVIDIA started by downloading from various academic and research sources a trio of datasets containing pointers to a curated set of millions of high-quality YouTube clips. NVIDIA then allegedly built and deployed a “mass download-and-ingest pipeline” to amass what one company executive allegedly described as “a video data factory that can yield a human lifetime visual experience worth of training data per day.” In doing so, the complaint alleges, NVIDIA carefully bypassed YouTube’s anti-scraping defenses, including the risk of YouTube blocking IP addresses used for mass downloading.

The lawsuit brings a sole count alleging that NVIDIA, in scraping videos from YouTube in this manner, bypassed technical protective measures controlling access to copyrighted audiovisual works in violation of Section 1201 of the Digital Millennium Act, codified at 17 U.S.C. § 1201. The lawsuit is brought on behalf of a putative class made up “[a]ll persons and entities in the United States who are creators and/or rights-holders of YouTube-hosted videos that [NVIDIA] accessed at the file level and acquired by scraping, downloading, or otherwise extracting the underlying video files from YouTube through circumvention of YouTube’s TPMs.”

The Case is No. 5:26-cv-00916.

Judge: Cousins, N.

Attorneys: William J. Edelman (Milberg, PLLC) for Devin Youngblood.

Companies: NVIDIA Corp.

News: Copyright TechnologyInternet CaliforniaNews GCNNews