A group of YouTubers filed a proposed class-action lawsuit against Snap on Friday in the U.S. District Court for the Central District of California, alleging the company scraped their videos without permission to train AI models for features like the app’s “Imagine Lens.”
The plaintiffs operate three YouTube channels with a combined 6.2 million subscribers. They claim Snap used their video content to develop AI systems, including the “Imagine Lens,” which enables users to edit images through text prompts. This action follows their earlier lawsuits against Nvidia, Meta, and ByteDance for comparable allegations of unauthorized use of YouTube videos in AI training processes.
The complaint targets Snap’s utilization of the large-scale video-language dataset HD-VILA-100M, along with other datasets created exclusively for academic and research purposes. These datasets incorporate videos from YouTube channels such as those owned by the plaintiffs. The suit details how Snap accessed this content despite restrictions designed to limit its application.
According to the plaintiffs, Snap bypassed YouTube’s technological measures, violated the platform’s terms of service, and disregarded licensing restrictions that explicitly bar commercial exploitation of the videos. These measures include protections against downloading or scraping content for purposes beyond personal, non-commercial viewing.
The lawsuit requests statutory damages for the alleged copyright infringement and a permanent injunction to prevent Snap from continuing such practices. This relief aims to compensate the creators and block future unauthorized training on their works.
Leading the case are the creators of the h3h3 YouTube channel, which has 5.52 million subscribers, and the golfing channels MrShortGame Golf and Golfholics. These channels produce content ranging from commentary and reactions on h3h3 to instructional golf videos on the others, all of which allegedly appear in the datasets Snap employed.
This filing joins a series of legal actions where content creators challenge AI model providers over copyright issues. Disputes have arisen from publishers, authors, newspapers, user-generated content sites, and artists. The nonprofit Copyright Alliance reports that more than 70 copyright infringement cases have been initiated against AI companies.
Court outcomes vary in these matters. A judge ruled in favor of Meta in its case against a group of authors. Anthropic reached a settlement with another group of authors, paying to resolve the claims. Numerous other cases remain in active litigation.
