Nvidia Illegally Trains Cosmos AI Software by Scraping YouTube Videos, Class Action Alleges

New to ClassAction.org? Read our Newswire Disclaimer

A proposed class action lawsuit alleges Nvidia has secretly and illegally scraped millions of YouTube videos to train its Cosmos artificial intelligence software at the expense of unknowing content creators.

Want to stay in the loop on class actions that matter to you? Sign up for ClassAction.org’s free weekly newsletter here.

The 11-page Nvidia lawsuit says the chip and software behemoth, reportedly valued at upward of $2 trillion, has reaped massive profits from its Cosmos AI deep learning software, which is critical for supporting other Nvidia products such as image generation, automated driving and the Omniverse platform used by millions of app and software developers worldwide.

The proposed class action says Nvidia’s scraping of YouTube videos without consent is a violation of the platform’s terms of service, which grant ownership rights to those who create video content uploaded to the website.

Deep learning is a variety of machine learning that utilizes artificial neural networks to learn from data, the filing explains, and the Cosmos service reportedly enables entities to “rapidly build, train, and deploy neural network models to address [] business demands.” In the field, deep learning applications can help with identifying objects and features in images, understanding the meaning of text, and converting text to images, the case adds.

The case says the aim of the Cosmos project was to “create a comprehensive video foundation model” that can support AI applications crucial to Nvidia’s other products. As the suit tells it, feeding video content, such as YouTube videos, to a deep learning service is essential to its development given that generative AI can learn from patterns in the videos.

The suit alleges Nvidia instructed employees to use an open-source YouTube video downloader and other tools to scrape thousands of videos from the platform “while evading detection and avoiding blocks by YouTube.” Nvidia needs tools to sidestep YouTube’s download blocks because using the platform’s content to train AI models amounts to a “clear violation” of its terms, the complaint states.

“Plaintiff and Class members did not consent to the use of their videos as training material for Cosmos,” the lawsuit asserts. “Nonetheless, their videos were downloaded for training Cosmos to be a critical support system for Nvidia products.”

The Nvidia lawsuit looks to cover all individuals or entities in the United States that uploaded any YouTube video fed to and used as training data for the Cosmos AI project without their consent.

Are you owed unclaimed settlement money? Check out our class action rebates page full of open class action settlements.