Three YouTube content creators, including the company behind H3H3 Productions, a solo golf presenter, and a golf channel, have filed a class action lawsuit in Seattle against Amazon. The lawsuit alleges that Amazon bypassed YouTube’s protections using virtual machines and rotating IP addresses to scrape videos without consent, utilizing them in training datasets for Nova Reel, a video AI model available through Amazon Bedrock. This lawsuit cites the anti-circumvention provisions of the Digital Millennium Copyright Act and follows similar legal actions the same group has filed against Nvidia, Meta, ByteDance, Snap, OpenAI, and Apple.
Ted Entertainment Inc., which operates H3H3 Productions and H3 Podcast Highlights on YouTube, alongside Matt Fisher of the MrShortGame Golf channel and Golfholics Inc., has filed the complaint in the US District Court for the Western District of Washington. Together, these plaintiffs represent over 2.6 million YouTube subscribers, around four billion total views, and more than 5,800 original videos. The suit names Amazon and targets Nova Reel as a product partially based on their content.
The lawsuit hinges on Section 1201 of the Digital Millennium Copyright Act, which prohibits bypassing technological protection measures set by copyright holders to control access to their works. The plaintiffs claim YouTube’s video protection systems are such measures and that Amazon deliberately circumvented these at scale to gather training data. A court siding with this theory could establish that downloading YouTube videos for AI training, even if publicly viewable, violates the DMCA by circumventing technical enforcement mechanisms.
The plaintiffs emphasize the permanent harm caused: “Once AI ingests content, that content is stored in its neural network and not capable of deletion or retraction.” They seek damages and injunctive relief, potentially forcing Amazon to cease distributing a model partially trained on their content or to retrain it without the contested material.
The complaint focuses on two academic datasets: HD-VILA-100M from Microsoft Research Asia and HD-VG-130M from Peking University and Microsoft. These datasets list YouTube URLs instead of the videos themselves. Legally, downloading the actual videos for AI training constitutes bypassing YouTube’s protections, which the plaintiffs claim Amazon did.
According to the complaint, Amazon did not just download videos. They used automated programs with virtual machines rotating IP addresses to avoid YouTube’s detection and blocking mechanisms. These combined technologies—automated mass downloading, virtual machine infrastructure, and IP rotation—are described as deliberate circumventions of YouTube’s technological protections. This pattern was claimed in a previous lawsuit against Nvidia, where similar technology was allegedly used to download 38.5 million video URLs.
Nova Reel is Amazon’s text-to-video generative AI model, introduced in December 2024 via Amazon Bedrock. It processes text prompts and images, creating video clips with a watermark for authenticity. As part of the larger Nova model family, it reflects Amazon’s competition acceleration in the enterprise AI sector, aiming to match rivals like Sora and Google Veo. Amazon’s extensive investment in AI infrastructure, including its collaboration with Uber for Trainium chips, underscores its wide-reaching ambitions in the AI stack, from cloud computing to generative media. The competitive drive for acquiring large-scale training data is underscored by SoftBank’s hefty $40 billion loan to OpenAI, highlighting the scale and intensity of investment in generative AI competition.
These plaintiffs have prior litigation experience, especially noteworthy in 2025 when AI training data practices became a coordinated legal focus. In December 2025, they initiated a class action against Nvidia in California, alleging similar use of academic datasets and infrastructure for video model training. In January 2026, lawsuits were filed against Meta, ByteDance, and Snap, with parallel lawsuits against OpenAI and Apple filed in April 2026 in California. The Amazon lawsuit, in Seattle, is the latest in this pattern.
Patent legal challenges against AI developers are increasing, with US copyright cases against AI companies surpassing 100, such as a lawsuit from Encyclopaedia Britannica and Merriam-Webster against OpenAI for unauthorized use of nearly 100,000 of Britannica’s articles as training data. The plaintiffs here aim to challenge the reliance on academic datasets for training AI, arguing that downloading videos using such datasets constitutes a DMCA violation, targeting the discrepancy between academic URLs and the necessary extraction. The outcome could mean that AI developers using academic video URL datasets face similar legal risks to those who directly download video footage.
Amazon, like other defendants in these suits, has not publicly commented on the lawsuit.