A Win for Visual Artists in the Copyright Battle Against AI Art Generators
By Daniel Jason Ain and Ariana Zhao
Tensions between creators and artificial intelligence (“AI”) companies continue to run high, fueling ongoing litigation.[1] In the latest win for creators, a class of visual artists survived a motion to dismiss their copyright infringement claims.[2]
Last year, our colleagues wrote about the copyright suit brought by a group of visual artists against tech companies on the basis of their art works being used to train AI tool Stability Diffusion without their consent. The text-to-image AI tool Stability Diffusion, created by Stability AI, one of the defendants in the case, is able to generate hyper-realistic images in response to text prompts given to it. The plaintiffs contend that Stability Diffusion was designed to produce new artwork by analyzing and synthesizing existing original art, effectively using these works as fodder for AI-generated creations, thus infringing on their copyrights. The artists additionally claim that because the works produced by Stability Diffusion are themselves infringing, third-parties that makes use of the tool are induced by design to infringe on their copyrights. Other defendants in the case, including research lab Midjourney and online art platform DeviantArt, used Stability Diffusion to train their own AI tools.
In December 2023, Senior District Judge William H. Orrick of the Northern District of California dismissed certain of the plaintiffs’ copyright infringement claims, but gave them leave to amend their complaint and cure the deficiencies identified in the Order. The plaintiffs amended the complaint to allege additional facts, including regarding how billions of images are able to be compressed and fed to the AI tool. The defendants once again filed a motion to dismiss, and this time around, additional claims survived.
Important in this case is plaintiffs’ claim that the output of these AI tools can recreate the works used to train the Stability Diffusion model itself. The plaintiffs point to Stability’s CEO’s own words, “We took 100,000 gigabytes of images and compressed it to a two-gigabyte file that can recreate any of those [images] and iterations of those”[3] in arguing that Stability “intended to foster infringement.”[4] Judge Orrick found plaintiffs’ arguments sufficient, noting that “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.[5] While allowing certain copyright infringements claims to survive, Judge Orrick dismissed the plaintiffs’ claims of unjust enrichment (with leave to amend), breach of contract, and breach of the implied covenant of good faith and fair dealing with prejudice.
The court’s ruling, aside from opening the floodgates to closer scrutiny of the inputs that are fed to AI visual art generators, underscores a growing recognition of the complexities introduced by AI. It remains clear that maintaining intellectual property protections in an era where AI plays an increasingly significant role in creative processes is crucial to preserving the production of original works and safeguarding the artists behind them. As AI continues to transform creative industries and copyright law evolves to adapt to the challenges that come with it, companies will need to ensure that their quest for innovation and commercial success do not infringe on creators’ rights.
This article is intended as a general discussion of these issues only and is not to be considered legal advice or relied upon. For more information, please contact RPJ Counsel Daniel Jason Ain who counsels clients in areas of entertainment, media and literary, intellectual property and employment law. Mr. Ain is admitted to practice law in the State of New York and the District of Columbia.