Several well-known YouTube creators are taking on one of Seattle’s biggest corporate residents in its own backyard, accusing Amazon of quietly raiding their videos to build a flashy new AI tool.
In a proposed class-action lawsuit filed Friday in federal court in Seattle, the creators say Amazon scraped millions of copyrighted YouTube videos to train Nova Reel, a text-to-video model the complaint says is offered through Amazon Bedrock. The named plaintiffs are Ted Entertainment (which runs h3h3 productions and h3 podcast highlights), Matt Fisher (known for the MrShortGame channel), and the Golfholics channel. They are asking for statutory damages, attorneys' fees, litigation costs, and a court order blocking the alleged conduct.
According to Law360, the complaint filed in the U.S. District Court for the Western District of Washington claims Amazon used automated downloading and descrambling tools, rotating IP addresses, and virtual machines to avoid detection while harvesting videos without permission. The suit argues that YouTube only allows controlled streaming and does not give ordinary users access to the raw video files, and that the alleged scraping violated both YouTube’s terms of service and federal copyright law.
As reported by KING 5, the lawsuit specifically calls out Amazon’s Nova Reel and alleges the company used the downloaded videos to feed, train, improve, and commercialize that model. An Amazon spokesperson told the outlet the company “does not comment on active litigation.”
What Is Nova Reel?
Amazon describes Nova Reel as a text-to-video model in its Bedrock lineup that turns prompts and reference images into short motion clips. The company’s model documentation says the Nova family, including Reel, is accessible through the Bedrock console. The creators’ complaint zeroes in on that product and alleges Amazon relied on scraped YouTube content as training data instead of properly licensed material.
What The Creators Want
Per Law360, the plaintiffs are asking the court for statutory damages, attorneys' fees, and an injunction that would bar Amazon from continuing the alleged data collection. The complaint also brings claims under the Digital Millennium Copyright Act’s anti-circumvention provisions, asserting that Amazon bypassed YouTube’s technological protections in order to access the underlying video files.
Legal Stakes And Wider Context
The case lands amid a wave of lawsuits testing how far tech companies can go in using copyrighted material to train generative AI models. Recent federal rulings have sent mixed messages for both creators and AI developers. As TechCrunch reported, high-profile decisions involving Anthropic and Meta showed courts can side with AI firms on narrow records, yet judges have also highlighted that outcomes often hinge on how training data was obtained and whether the use harms the market for the original works.
The Seattle case is still in its early stages. Next steps typically include motions to dismiss, discovery fights over how Amazon gathered data, and potential battles over whether the suit can proceed as a class action. However it shakes out, the litigation could influence how the industry handles licensing and the use of user-generated video as training fuel for new audio-visual generative models.
