Offit Kurman Blogs

On October 2, 2026, a Northern California District Judge denied Chinese AI giant ByteDance’smotion to dismiss in Ted Entertainment, Inc. v. ByteDance Inc., a class action by several YouTube content creators who allege their videos were scraped to train ByteDance’s text-to-video models, MagicVideo and Seedance. The case is unusual because it pleads no copyright infringement claim at all. The sole count arises under Section 1201(a) of the Digital Millennium Copyright Act, which prohibits circumventing a technological measure that effectively controls access to a copyrighted work. That framing sidesteps the fair use debate over AI training and asks a narrower question about how the videos were obtained. The plaintiffs allege ByteDance’s researchers downloaded videos directly from YouTube, bypassing the video platform’s protective measures.

ByteDance raised two defenses. First, it argued the plaintiffs lacked standing because their harms flow from copying and training rather than from circumvention, and because the private right of action belongs to the owner of the measure, which is YouTube. Second, it argued that videos free to watch cannot be “access controlled,” since YouTube’s measures only restrict copying. The court disagreed on each point and found the lost per-view advertising revenue traceable to circumvention, since scraped retrievals register no views. It held that copyright owners are squarely within the class Congress authorized to sue. On the merits, the court relied on the Ninth Circuit’s decision in Disney v. VidAngel for the principle that one measure can serve as both an access control and a copy control. Since the plaintiffs allege that proof-of-origin tokens are required even to stream a video, the court found it plausible that these measures control access at the outset. The court adopted the view that these barriers may be modest, but they remain barriers.

This ruling adds to a growing line of decisions treating platform security as a legal protection that travels with the creator’s content, and the same plaintiffs have filed parallel suits against several other major technology companies. For creators, studios, and brands that publish online, the practical point is how and where you distribute your work shapes the claims available to you later. Choosing platforms with real access controls, keeping records of the revenue tied to views, and writing license terms that prohibit bulk extraction can each support a claim that does not depend on winning the fair use argument over AI training.