Artists Copyright Claims Against AI Image Generators Advance in Court

Artists taking on major AI image generator platforms have scored a significant legal victory, as a federal judge allowed their copyright infringement claims to move forward. 

The ruling suggests that AI models like Stable Diffusion may infringe copyrights "by design" through their very operation.

Cartoonist Sarah Andersen, illustrator Kelly McKernan and concept artist Karla Ortiz sued Stability AI, the startup commercializing Stable Diffusion, Midjourney and DeviantArt, alleging their works had been used to train their generation models without permission.

The trio alleges that the illicit use of their works in the AI generation of new images resulted in the copying of the original works.

In the latest ruling, a Northern California District Court judge allowed the claims of copyright infringement to proceed as models like Stable Diffusion are built on works and by sheer operation, perpetuate copyright infringement.

“This is a case where plaintiffs allege that Stable Diffusion is built to a significant extent on copyrighted works and that the way the product operates necessarily invokes copies or protected elements of those works,” wrote District Judge William Orrick.

“The plausible inferences at this juncture are that Stable Diffusion by operation by end users creates copyright infringement and was created to facilitate that infringement by design … the allegations of induced infringement are sufficient.”

Related:Stability AI Wins Partial Legal Victory Over Infringement Claims

The ruling is significant as it provides the plaintiff with the ability to demand information on how the image generation platforms obtained training materials for their AI models as part of discovery.

Stability, along with other image generation vendors, used the LAION-5B dataset, an open source corpus of billions of images scraped from the internet to train models like Stable Diffusion.

Judge Orrick said the artists’ copyright claims could proceed as it was “undisputed” that their protected works were included in the LAION dataset.

The AI companies tried to stop the artists from getting broader access to the dataset and the images it contains. They argued that the artists needed to specify which individual images they thought were included in LAION. 

However, the judge supported the artists and ruled that their claims were believable enough to show that their work was indeed part of the dataset.

“Given the unique facts of this case, including the size of the LAION datasets and the nature of defendants’ products… that level of detail is not required for plaintiffs to state their claims.

Related:Tech Lobbying Group Launches Campaign for AI Fair Use Rights

The case has been ongoing since January 2023, with the district court handing the image-generation platforms a partial win last December by ruling that two of the plaintiffs could not bring infringement claims as they failed to register their works with the U.S. Copyright Office. However, the court did allow the artists to file an amended complaint.

There were some wins in the latest ruling for the AI startups, however, as the judge dismissed claims that they violated the Digital Millennium Copyright Act (DMCA) by allegedly removing information identifying owners of the disputed works.

Also dismissed were the artist’s claims that the AI companies unjustly enriched themselves through the use of their works - meaning the case now solely boils down to copyright infringement claims. However, the artists were given another attempt to amend their claims of unjust enrichment.

The case is one of a growing number brought by artists against generative AI platforms.

The lawyers representing the artists in this case have filed a class action lawsuit against Google, citing similar claims of infringement. They argue that Google's Imagen models were trained on a smaller version of the LAION dataset, LAION-400M, which includes their works. As a result, the models allegedly generate copies of their work.