YouTubers sue Amazon over scraping videos to train Nova Reel AI
2026年4月3日 · 自動從公開來源蒐集,未經人工審核。
這篇內容目前只有 英文 原文版本。
What happened
On April 3, 2026, three YouTube creators filed a class action in the U.S. District Court for the Western District of Washington (Seattle) against Amazon, alleging the company illegally used their YouTube videos to train and improve its Nova Reel generative AI video model. Named plaintiffs include Ted Entertainment, Inc. (Ethan and Hila Klein's h3h3 Productions/H3 Podcast, 5,800+ videos, 4B+ views), Matt Fisher (MrShortGame Golf, 500K+ subscribers), and the Golfholics channel. The complaint alleges Amazon used datasets earmarked for academic use only, circumvented YouTube's copyright-protection measures (DMCA claim), and deployed automated download tools on virtual machines cycling IP addresses to extract millions of videos. Plaintiffs seek damages, restitution, and injunctive relief; the suit argues creators have no way to recover IP once ingested — "once AI ingests content, that content is stored in its neural network and not capable of deletion or retraction." Amazon declined comment.
Why it matters to creators
Video creators face the same training-data fight as authors and newsrooms, but with the added DMCA anti-circumvention angle: the suit alleges Amazon deliberately defeated technical protections to get the footage. If video-generation models can be trained by scraping YouTube despite its terms and TPMs, every video creator is effectively an uncompensated training-data supplier for models that will compete with them.
What is not yet known
The case is in litigation; no court has ruled on the DMCA-circumvention claim or whether YouTube's terms of service bar this kind of scraping for AI training.
來源
- YouTubers sue Amazon over scraping videos to train Nova Reel AI — GeekWire (secondary)