YouTube creators sue Amazon over training Nova Reel on their videos

Three YouTube creators have filed a class action against Amazon in federal court in Seattle. They allege the company downloaded their videos from YouTube without permission and used the material to train Nova Reel, its own video generation model available through Amazon Bedrock.

Who filed the suit

The plaintiffs are Ted Entertainment, the company behind the channels h3h3 Productions and H3 Podcast Highlights, along with Matt Fisher, creator of the golf channel MrShortGame Golf, and the Golfholics team. According to the complaint, together they have more than 2.6 million subscribers, about 4 billion views and over 5,800 original videos on YouTube.

The suit was filed on April 3, 2026, in the U.S. District Court for the Western District of Washington. Formally, it is an attempt to turn the case into a class action on behalf of a broader group of content creators. The plaintiffs are seeking compensation, disgorgement of unlawfully obtained profits and an injunction that could affect further distribution of the model.

"If such circumvention goes unpunished, creators will publish their work on

YouTube less often."

How the scraping is described

The central claim rests on Section 1201 of the DMCA, which covers circumvention of technical protection measures. The creators are not disputing merely that public videos were viewed. Their position is harder-edged: Amazon allegedly used automated download tools, virtual machines and IP address rotation to pull video files in bulk and get around YouTube's restrictions.

The complaint separately mentions the academic datasets HD-VILA-100M and HD-VG-130M. They contain not the videos themselves but links to YouTube videos. According to the plaintiffs, to turn such URL lists into training data for a model, Amazon must have downloaded the actual files from the platform, and it is this step that, in their view, violates the law and the terms of service.

• The plaintiffs argue that YouTube already has technical barriers against bulk video extraction.
• Amazon, according to the suit, bypassed those barriers using automated infrastructure.
• The dispute is not only about copyright but also about how the data was obtained.
• A separate demand is a ban on using the model if it was trained on the disputed video set.

Another important argument from the plaintiffs concerns the irreversibility of training. They claim that once content has entered a model, it cannot simply be "deleted back out," so the harm is not limited to the single act of downloading. That makes the case awkward for Amazon: even if the disputed dataset is no longer in use, questions about the already trained version of Nova Reel do not go away.

Why it matters for Amazon

Nova Reel is Amazon's video model, introduced in December 2024 and available through Bedrock. According to AWS documentation, it can generate videos from text and images and supports clips from 6 seconds to 2 minutes long, while version 1.1 added a multi-shot mode and better style consistency between scenes. For Amazon, this is not a side experiment but part of a big bet on enterprise generative AI.

Against that backdrop, the suit hits more than the company's reputation. It puts in question the provenance of the data the model may have been trained on. If the court accepts the plaintiffs' logic, the risk would extend beyond Amazon to any company that took academic collections of YouTube links and then downloaded the source videos in bulk to train video models.

It is also part of a broader series of suits. The same plaintiffs have already filed similar complaints against Nvidia, Meta*, ByteDance, Snap, OpenAI and Apple. The overall strategy is clear: target not abstract training on the open internet but a specific technical mechanism through which companies allegedly bypassed a platform's protections and turned publicly available videos into a commercial AI asset.

What this means

This case could become one of the most telling disputes over video data for AI. If the court agrees that bulk downloading of YouTube videos for model training violates the DMCA even when the content is publicly accessible, developers will have to rethink not only their datasets but the entire data collection pipeline for generative video.

*Meta has been designated an extremist organization and is banned in Russia.