Background
Visual artists including Sarah Andersen, Kelly McKernan and Karla Ortiz filed a putative class action against Stability AI Ltd., Midjourney Inc. and DeviantArt Inc. alleging that the defendants used their copyrighted works to train AI image-generation models such as Stable Diffusion without permission, and that the models produce new images in the artists’ styles. The court denied in part and granted in part motions to dismiss, allowing core direct copyright infringement and inducement claims against Stability AI to proceed while dismissing or narrowing other claims.
AI interaction
The court held that plaintiffs sufficiently alleged that “Stable Diffusion by operation by end users creates copyright infringement and was created to facilitate that infringement by design.” The decision is significant for placing algorithmic training of image-generation models, rather than just user outputs, at the heart of copyright liability questions. It signals that AI companies may be held accountable not only for outputs but for the training practices and dataset inclusion of copyrighted works.
Note: The ruling advances the legal frontier on AI-training dataset accountability, allowing direct and inducement claims to survive a motion to dismiss and placing pressure on generative-AI businesses to clarify data-ingestion practices.