Judge signals ByteDance unlikely to dismiss YouTubers' scraping suit over AI video-model training

What happened

A federal judge in San Francisco signaled she will deny ByteDance's bid to dismiss a class action by YouTubers (Ted Entertainment, Inc. et al. v. ByteDance, N.D. Cal., No. 3:25-cv-10933) accusing the TikTok parent of scraping millions of YouTube videos to train its text-to-video "MagicVideo" AI model. After a Thursday hearing, Judge Jacqueline Scott Corley indicated ByteDance's motion to dismiss is unlikely to succeed, according to Bloomberg Tax reporting.

ByteDance argued the creators' Digital Millennium Copyright Act claims fail because downloading publicly available videos does not qualify as the "unauthorized access" barred by the DMCA's anti-circumvention provision. Judge Corley pushed back: "I don't know how that's not, at this stage, circumventing," she said. The complaint alleges ByteDance used automated tools to evade YouTube's technical protection measures, downloaded and stored the videos, and assembled training corpora for its AI models, with each act of circumvention constituting a separate DMCA §1201 violation.

Why it matters to creators

Video creators whose work was posted on YouTube are the plaintiffs here — this is directly about how a major AI company's training pipeline may have treated uploaded creator videos, including bypassing platform access controls to assemble training data. If the DMCA §1201 theory survives, it gives creators a separate legal lever beyond ordinary copyright claims against large-scale video scraping for model training.

What is not yet known

The judge's comments are a hearing signal, not a final ruling on the motion — the case would proceed to discovery if the motion is denied. Bloomberg's full reporting is paywalled; the precise reasoning will only be clear when a written order issues.