Northern California – In a landmark decision, a federal judge has permitted artists to advance their copyright infringement claims against major AI image generation platforms. The ruling, delivered by District Judge William Orrick, suggests that AI models such as Stable Diffusion may infringe on copyrights "by design" through their operational mechanics.
Cartoonist Sarah Andersen, illustrator Kelly McKernan, and concept artist Karla Ortiz have filed a lawsuit against Stability AI, the developer behind Stable Diffusion, as well as against Midjourney and DeviantArt. The artists claim their works were used without permission to train the AI models that generate new images.
The plaintiffs argue that the use of their copyrighted works in training these AI systems resulted in unauthorized reproductions of their original creations. Judge Orrick’s ruling indicates that the claims are substantial enough to warrant further examination, particularly because the operation of AI models like Stable Diffusion appears to inherently involve the reproduction of copyrighted content.
“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,” Judge Orrick stated. “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.”
The ruling is a significant development as it grants the artists the ability to demand detailed information about how these AI platforms sourced their training materials. Stability AI and other companies involved used the LAION-5B dataset, a vast collection of images gathered from the internet, to train their models. Judge Orrick noted that it is “undisputed” that the plaintiffs’ works were included in the LAION dataset.
In response to the artists' push for broader access to this dataset, the AI companies had argued that the artists should specify which images they believed were used. However, Judge Orrick sided with the artists, ruling that given the extensive nature of the dataset and the specifics of the case, such detailed identification was not required at this stage.
While the ruling is a victory for the artists, it is not without limitations. The judge dismissed claims that the AI companies violated the Digital Millennium Copyright Act (DMCA) by allegedly removing information about the owners of the disputed works. Additionally, claims of unjust enrichment were also dismissed, narrowing the case to focus solely on copyright infringement. However, the artists have been given another chance to amend their unjust enrichment claims.
This case, which began in January 2023, follows a partial legal victory for the AI companies in December 2023, when the court ruled that two of the plaintiffs could not pursue infringement claims due to their failure to register their works with the U.S. Copyright Office. Nevertheless, the court allowed the artists to file an amended complaint.
The dispute is part of a growing wave of legal actions by artists against generative AI platforms. Lawyers representing the plaintiffs have also filed a class action lawsuit against Google, alleging similar claims of copyright infringement involving Google's Imagen models and a smaller subset of the LAION dataset, LAION-400M.
