Andersen v Stability AI Ltd is a proposed class action brought by visual artists - lead plaintiff Sarah Andersen, joined by Kelly McKernan, Karla Ortiz and others - in the United States District Court for the Northern District of California, Case No. 23-cv-00201-WHO, filed in January 2023. The artists allege that Stability AI and other defendants, including Midjourney, DeviantArt and Runway, built image-generating models by training on their works without permission. It was one of the first major copyright suits aimed at generative-AI image tools.
On August 12, 2024, Judge William H. Orrick issued an order on the defendants’ motions to dismiss the amended complaint, granting it in part and denying it in part (2024 WL 3823234). Critically for the plaintiffs, the judge allowed their core copyright theories to proceed. The order let stand the artists’ direct copyright-infringement and induced-infringement claims against Stability AI, reasoning that whether the company was liable turned on “how Stable Diffusion works” - a factual question “better addressed on summary judgment, after discovery.” Some claims were narrowed or dismissed, including a claim under section 1202 of the Digital Millennium Copyright Act over copyright-management information.
The ruling did not decide whether training on the artists’ works was infringing or fair use; it decided only that the allegations were plausible enough to move into discovery. As of the verified date the case remained ongoing.
Why business readers should care: Andersen was an early signal that courts would let generative-AI training claims survive past the pleading stage and into fact discovery rather than throwing them out at the threshold. By tying liability to how the model actually works, the order made the internal mechanics of training and generation - normally trade secrets - central to the litigation, a pattern echoed in later cases such as 2025-bartz-v-anthropic and 2025-getty-v-stability-ai-uk.