Unreported / Non-Citable
Background
Digital artist Austin Beaulier alleged that Roblox used copies of his three-dimensional models from a large research dataset to train generative-AI systems. His Sketchfab models carried attribution and licensing information, including a NoAI tag, which he characterized as copyright management information, or CMI.
Beaulier brought putative class claims under section 1202(b) of the Digital Millennium Copyright Act. He alleged that Roblox stripped CMI while preprocessing the files and then distributed CMI-free representations through its AI platform. Roblox moved to dismiss, arguing that the complaint did not adequately plead removal, intent, distribution, or the statute’s knowledge requirements.
The Court’s Holding
The court found that Beaulier had standing because impairment of his asserted ability to control, license, and monetize his own works was a concrete injury plausibly traceable to Roblox. But standing was not enough to state a DMCA claim.
The removal claim did not plausibly allege an affirmative, intentional removal of CMI rather than an incidental consequence of converting files for machine-learning use. The distribution claim also failed because internal use is not distribution, and the complaint’s general assertion that works were distributed through an AI platform did not allege facts showing a transfer of Beaulier’s own works. Those gaps also undermined the required allegations that Roblox knew its conduct would facilitate or conceal infringement. Both counts were dismissed with leave to amend.
Key Takeaways
- A creator may establish standing by alleging that removal of CMI impaired the creator’s ability to control and monetize identified works.
- DMCA section 1202 requires facts showing an affirmative and intentional removal of CMI; loss of metadata during routine format conversion may not suffice.
- Internal use of a work for AI training is not, by itself, distribution under section 1202(b)(3).
- Plaintiffs must connect the defendant’s knowledge to identifiable infringement, not merely invoke the universal possibility that missing CMI could encourage infringement.
Why It Matters
The order supplies a practical pleading roadmap for artists and AI companies facing DMCA claims over training data. Creators need technical facts about how metadata or NoAI signals were removed, who made that choice, and how copies of their particular works moved beyond internal systems.
For platforms, the decision underscores that standing and statutory liability are separate questions. A creator can allege a real economic injury yet still lose at the pleading stage if the complaint does not tie the processing pipeline to intentional removal, external distribution, and the DMCA’s demanding knowledge standard.