YouTuber sues Nvidia over AI training with his videos

In the USA, a YouTuber is seeking a class action lawsuit because his videos were allegedly used for AI training. However, this is not about copyright.

Save to Pocket listen Print view
Nvidia with greening

(Image: Michael Vi/Shutterstock.com)

2 min. read
By
  • Nico Ernst

After OpenAI, Nvidia is now also being sued for the alleged use of YouTube videos as training data for its own artificial intelligence. The lawyers of David Millette, a previously unknown YouTuber, are making a new type of allegation that is not based on any copyright infringements.

As the specialist medium Legal Dive reports , Millette's lawyers filed the lawsuit on August 14, 2024, two weeks after the man filed a similar lawsuit against OpenAI. In separate lawsuits, the YouTuber is seeking an injunction and damages. The filings with a Californian court, which have not yet been decided on, are structured as class action suits. Other affected parties can therefore join them.

According to the statement of claim (PDF), the plaintiff's lawyers accuse Nvidia of having enriched itself with prohibited business practices at the expense of the YouTuber. Literally, the methods are said to be "unfair, immoral, unethical, oppressive, unscrupulous or harmful to consumers". This is noteworthy because it represents a new approach in the way creators defend themselves against AI training with their works. So far, other cases have not yet conclusively clarified whether unsolicited training with copyrighted content is legally permissible. According to Legal Dive, none of David Millette's lawsuits raise allegations under US copyright law. Unlike German or European copyright law, this primarily regulates rights of use.

The lawsuits come shortly after the US service "404 Media" uncovered the background to Nvidia's "Cosmos" project because of leaked internal documents and statements from former employees who were trying to remain anonymous. According to this, the company is said to have endeavored to automatically download millions of hours of videos from YouTube to turn them into training material for its own AIs. Among other things, servers from third-party providers are said to have been rented to conceal Nvidia's IP addresses. According to the report, the company argued internally to employees who had legal and ethical concerns about "fair use" of the data, among other things.

(nie)

Don't miss any news – follow us on Facebook, LinkedIn or Mastodon.

This article was originally published in German. It was translated with technical assistance and editorially reviewed before publication.