California Federal Court Trims Lawsuit Against Stability AI, Midjourney, and DeviantArt in Generative AI Artwork Case

Judge William Orrick in the Northern District of California has dismissed some claims in a lawsuit brought by visual artists Sarah Andersen, Kelly McKernan, and Karla Ortiz against Stability AI, Midjourney, and DeviantArt. The artists allege that the companies misused the artist’s copyrighted works in connection with the companies’ generative artificial intelligence systems and products. Judge Orrick dismissed all allegations against Midjourney and DeviantArt, but gave Plaintiffs leave to amend their complaint. Judge Orrick also dismissed copyright infringement claims brought by McKernan, Ortiz, and others in the artists’ class who had not obtained copyright registrations, a prerequisite to filing a copyright claim, but is allowing Andersen to pursue her claim that Stability AI's use of her work to train Stable Diffusion infringed her copyrights.

The artists’ class action, Andersen et al. v. Stability AI LTD et al., N.D. Cal. 3:23-cv-00201 (Filed Jan. 13, 2023), argues that the companies’ use of copyrighted works to train AI models, and the resulting image outputs, infringed the artists’ copyrights in those works.  The artists complain that these allegedly infringing activities are taking place without obtaining consent from the artists or the publishers of the artworks. The original complaint alleges direct copyright infringement by downloading, storing, and creating derivatives of copyrighted works of artists for training AI models; vicarious copyright infringement by enabling consumers to use the artists’ images through AI programs to generate derivative works; DMCA violations; right of publicity violations; and unlawful competition. This case is one of many current copyright cases by artists and authors filed against owners of generative AI programs, expected to shape the law on the extent to which companies can scrape copyrighted artworks for training purposes, and the extent to which artists can successfully assert copyright infringement owing to the creation by AI products using artworks from the AI training models.

Judge Orrick noted in his ruling that even Stability AI recognizes that the copying Stable Diffusion is accused of cannot be resolved without further investigation into the similarity of the works. He said he was not convinced that the claims focused on the generative AI outputs could be successful without a showing of substantial similarity to the artists’ works, but allowed that the plaintiffs’ claims could be amended to address the court’s concerns. He also dismissed without prejudice all claims related to the right of publicity and unfair competition, possibly narrowing the scope of issues in subsequent generative AI lawsuits. The artists' attorneys plan to file an amended complaint next month to address the court's concerns about their other claims and noted that their "core claim" survived.

The filing of this case was noteworthy as it was one of the first lawsuits in an attempt by artists to curtail the use of generative AI during the initial boom of AI programs offered to consumers.  Similar challenges have been lodged against AI training models and related programs that are used to create code, photographic images, and books. During the 10 months since this case was filed, it has been under careful scrutiny, with the court taking a nuanced and firm look at the allegations to hone in on the core legal issues potentially affecting artist rights in the new world of AI powered works of authorship

Action Items:

• Artists and copyright owners should consult with legal counsel to ensure their copyrights are protected and to explore potential claims against companies that may be misusing their work in generative AI systems.
• Organizations utilizing generative AI systems should review their practices and procedures to ensure compliance with copyright laws and consider obtaining proper permissions or licenses for the use of copyrighted material.
• Organizations using vendors that use generative AI systems should consider whether current contractual arrangements permit such practices, and if so, whether any modifications are necessary to address the potential risk associated with generative AI systems.

We will continue to monitor developments in this area, and update our clients accordingly.