Midjourney, Stability AI and DeviantArt score in copyright case

The contentious issue of whether AI art generators violent copyright — since they are by and large trained on human artists’ work, in many cases without their direct affirmative consent, compensation, or even knowledge — has taken a step forward to being settled in the U.S. today.

U.S. District Court Judge William H. Orrick, of the Northern District of California, today filed a decision in a copyright infringement class action lawsuit brought against Stability AI (creator of the popular open-source Stable Diffusion text-to-image AI generator), Midjourney (another AI image generator based on Stable Diffusion) and popular image sharing service and social network DeviantArt (which released its own AI image generator based on Stable Diffusion, “DreamUp” back in late 2022). The lawsuit was filed by three artists —Sarah Anderson, Kelly McKernan, and Karla Ortiz.

Full disclosure: VentureBeat regularly uses Midjourney, Stable Diffusion, and other AI art image generators to create article header art and other art for our digital presence.

Motion to dismiss ‘largely granted’

The three AI image generator companies had filed a motion to dismiss the copyright infringement case against them by the artists, and today Judge Orrick largely granted it, writing “the Complaint is defective in numerous respects.”

Orrick spends the rest of his ruling explaining why he found the artists’ complaint defective, which includes various issues, but the big one being that two of the artists — McKernan and Ortiz, did not actually file copyrights on their art with the U.S. Copyright Office.

Also, Anderson copyrighted only 16 of the hundreds of works cited in the artists’ complaint. The artists had asserted that some of their images were included in the Large-scale Artificial Intelligence Open Network (LAION) open source database of billions of images created by computer scientist/machine learning (ML) researcher Christoph Schuhmann and collaborators, which all three AI art generator programs used to train.

Roar like a LAION

The size of the LAION database may be helpful in protecting the AI companies, as Orrick writes:

“The other problem for plaintiffs is that it is simply not plausible that every Training Image used to train Stable Diffusion was copyrighted (as opposed to copyrightable), or that all DeviantArt users’ Output Images rely upon (theoretically) copyrighted Training Images, and therefore all Output images are derivative images.

Even if that clarity is provided and even if plaintiffs narrow their allegations to limit them to Output Images that draw upon Training Images based upon copyrighted images, I am not convinced that copyright claims based a derivative theory can survive absent ‘substantial similarity’ type allegations. The cases plaintiffs rely on appear to recognize that the alleged infringer’s derivative work must still bear some similarity to the original work or contain the protected elements of the original work.”

In other words — because AI image generators reference art by many different artists when generating new imagery, unless it is possible to prove that the resulting image referenced solely or primarily copyrighted art, and is substantially similar to that original copyrighted work, it is likely not infringing of the original work.

The fight continues…

Yet, Orrick does invite the artists to amend their claims and refile a narrower lawsuit citing specifically infringed copyrighted images.

The judge also allowed one count — for direct copyright infringement against Stability AI for copying Anderson’s 16 copyrighted works without authorization — to move forward. Read the full ruling document below (via Aaron Moss):