Three YouTube creators — H3H3 Productions, MrShortGame Golf, and Golfholics — have filed a proposed class action lawsuit against Amazon in Seattle, alleging the company scraped their videos without consent to train its Nova Reel text-to-video AI model. The complaint claims Amazon used virtual machines and rotating IP addresses to bypass YouTube's technical protections, invoking the DMCA's anti-circumvention provisions (Section 1201). The scraping allegedly relied on two academic datasets (HD-VILA-100M and HD-VG-130M) that contain YouTube video URLs, which Amazon then used to download actual video files. The same legal team has filed similar suits against Nvidia, Meta, ByteDance, Snap, OpenAI, and Apple. The suits collectively challenge the practice of using academic URL datasets as a legal cover for mass video extraction, and if courts accept the theory, AI developers using such datasets could face broad DMCA liability.
Table of contents
The complaint and its legal theoryHow the scraping allegedly workedNova Reel and Amazon’s video AI ambitionsA pattern of lawsuits, and a legal theory in developmentSort: