Judge lets core copyright claims against Stability AI proceed in Andersen
Judge William Orrick dismissed DMCA claims but allowed direct copyright-infringement and inducement claims against Stability AI, Midjourney and DeviantArt to proceed to discovery.
- Courts & copyright
- Notable
Judge William Orrick of the US District Court for the Northern District of California ruled on motions to dismiss in Andersen v. Stability AI, the class action brought by illustrators Sarah Andersen, Kelly McKernan and Karla Ortiz against Stability AI, Midjourney, DeviantArt and Runway AI over the training of Stable Diffusion and related image generators. The order let the case’s central copyright claims move forward to discovery for the first time since the artists filed suit in January 2023.
Orrick dismissed the plaintiffs’ claims under Section 1202 of the DMCA — which prohibits removing or falsifying copyright management information — with prejudice against Stability AI, and rejected a breach-of-contract claim against DeviantArt outright. But he denied the defendants’ motions to dismiss the core claims: direct copyright infringement and inducement of infringement against Stability AI, direct infringement against Runway AI under both a “model theory” (that the trained model itself embeds copyrighted works) and a distribution theory, and a Lanham Act false-endorsement and trade-dress claim against Midjourney. The court found the artists had plausibly alleged that Stable Diffusion was built “to a significant extent on copyrighted works” and rejected Stability AI’s argument that storing works only in algorithmic, rather than literal, form defeated an infringement claim at the pleading stage.
The ruling was not a verdict on the merits — it only found the artists’ theories legally sufficient to survive dismissal — but it was consequential because it kept alive the “model theory” argument that the AI industry had hoped courts would reject early: that a trained model can itself constitute an infringing copy of the works used to train it, independent of what it later generates. Trial was subsequently set for September 2026. The case ran in parallel with Getty Images’ separate suit against Stability AI, which Getty refiled in California in August 2025 after its UK claim narrowed.