Background
Ace Cam, which operates the Random Golf Club YouTube channel, and Businessing, which operates the Ali Spagnola channel, brought a proposed class action against generative-video company Runway AI. They allege that Runway obtained their YouTube videos without authorization and used the files to train and improve its artificial-intelligence video systems.
The creators did not frame the surviving claim as a conventional copyright-infringement dispute over whether AI training is fair use. Instead, they invoked Section 1201(a) of the Digital Millennium Copyright Act, which prohibits circumventing a technological measure that effectively controls access to a copyrighted work. Their complaint describes YouTube protections including a rolling cipher, rate limits and IP blocking, session-bound URLs, CAPTCHA challenges, and proof-of-origin tokens.
According to the complaint, Runway or its agents bypassed those measures through automated download tools and rotating proxy addresses. Runway moved to dismiss, arguing principally that publicly viewable YouTube videos were not access-controlled works, that the cited measures regulated copying rather than access, and that the creators themselves had not installed the protections.
The Court’s Holding
Judge Lewis A. Kaplan denied Runway’s motion to dismiss. At the pleading stage, the court had to accept the complaint’s well-pleaded factual allegations as true and decide only whether they stated a legally plausible claim. The ruling therefore does not establish that Runway actually circumvented YouTube’s systems, infringed any copyright, or owes damages.
The court held that public availability did not necessarily defeat the Section 1201 claim. Viewers may watch the videos for free, but the complaint plausibly alleged that YouTube still controls access to the underlying video files through an authorized playback process. A technological protection measure also may regulate both access and copying; the DMCA does not require a court to put a dual-purpose safeguard into only one category.
The creators also plausibly alleged circumvention. Their factual theory was that Runway continued obtaining file-level copies despite protections designed to limit automated downloading, including session restrictions, ciphers, rate limits, and bot-detection measures. Finally, the court rejected the argument that the copyright owners had to deploy the protections personally. The complaint plausibly alleged that the creators authorized YouTube to apply those safeguards when they uploaded their works.
Key Takeaways
- A work can be freely viewable while the technical pathway to its underlying file remains protected by an access control under the DMCA.
- A technological measure may control both access and copying; that overlap does not automatically remove it from Section 1201(a).
- Copyright owners can plausibly rely on platform-deployed protections when they authorized the platform to safeguard uploaded works.
- The decision addresses pleading sufficiency only. It does not resolve liability, class certification, damages, or the broader fair-use question for AI training.
Why It Matters
The opinion highlights a legal risk for AI developers that is distinct from the much-debated question whether copying works for model training is fair use. Even if a developer ultimately has a copyright defense concerning training, the method used to acquire source material may create separate exposure under the DMCA’s anti-circumvention rules.
For creators and online platforms, the ruling suggests that public streaming does not necessarily amount to permission to extract raw files at scale. For AI companies, it reinforces the practical importance of documenting data provenance, honoring platform access controls, and separating authorized viewing from automated downloading. The factual allegations remain unproven, but the creators may now proceed into discovery.
Surfaced via Law360 IP.