Artists Secure Major Victory in Landmark Intellectual Property Case Against AI-Generated Content Companies

On August 12, 2024, Judge William H. Orrick of the Northern District of California issued a landmark 33-page ruling in Andersen et al v. Stability AI Ltd., 23-cv-00201-WHO (N.D. Cal. Oct. 30, 2023) that could reshape the landscape of intellectual property law in the age of artificial intelligence (AI). Originally filed in February of 2023, the class-action lawsuit centers around the alleged unauthorized storage and use of various visual artists’ works to train AI models to create new content.

The plaintiffs include a group of ten artists whose works range from photography to the written word. The defendants include generative AI developers Stability AI, DeviantArt, Midjourney, and Runway AI. The lawsuit cleared initial obstacles when Judge Orrick allowed direct and contributory copyright and trademark infringement claims to progress. The Judge did, however, grant the defendants’ motion to dismiss other claims, including those addressing breach of contract, breach of the implied covenant of good faith and fair dealing, unjust enrichment, and breach of the Digital Millennium Copyright Act (DMCA).

Pursuant to the plaintiffs’ complaint, the defendants scraped the Internet for content without permission and proceeded to store and aggregate it into very specific datasets that were then used to train their AI systems to create new, “original” content based on the works. The AI systems include, by way of example, DeviantArt’s Midjourney and DreamUp, and Stability AI’s Stable Diffusion and DreamStudio. Plaintiffs contend that their intellectual property rights, as protected by the U.S. Copyright Act of 1976 and the Lanham Act, were violated by the defendants who were unjustly enriched at their expense. The plaintiffs requested monetary damages and injunctive relief. The defendants, however, argued that their actions were protected under the fair use doctrine, which allows for the limited exploitation of copyrighted material without prior permission from the applicable rights holder. The defendants also contended that their algorithmic models were created to produce new, original works of art that were separate and distinct from the materials created by the artists themselves. They also held that the plaintiffs’ copyright claims should fail as a matter of law since they did not specifically identify each allegedly infringed work of art.

After initially dismissing the plaintiffs’ claims for lack of clarity in October of 2023, Judge Orrick held that their amended complaint delineating the alleged infringement was sufficient to proceed, stating that the software programs used by the AI companies were “built to a significant extent on copyrighted works” and “necessarily invoke” elements of copies of such works. He went on to state that the programs were “created to facilitate that infringement by design.” Judge Orrick also reasoned that given the enormous size of the datasets, the plaintiffs were not required to include in their complaint a detailed listing of the infringed content to survive a motion to dismiss. With the court permitting the infringement claims to proceed, attention has now shifted to each side’s detailed arguments and supporting evidence. Per Judge Orrick, “Whether plaintiffs will be able to prove their claims is a different matter and those claims will be tested on an evidentiary basis at summary judgment.”

Judge Orrick’s pivotal ruling affirms that intellectual property rights in AI-generated content remain with those individuals who created the underlying works in the first instance, and as such, any unauthorized use of such content would necessarily constitute infringement. Intellectual property law is inherently rooted in human creativity. Despite AI’s ability to generate new content, it cannot be considered an author under existing intellectual property law. Human creativity remains vital to the ownership and authorship of creative content. The Judge’s decision marks a critical moment and key reference point in the legal analysis surrounding intellectual property law and the rapidly expanding use of artificial intelligence and similar forms of technology. It underscores the need for AI companies to obtain proper permissions and licenses before incorporating pre-existing content in their datasets. The outcome will most likely influence how AI systems are developed and trained going forward. The ruling sets a new precedent respecting how AI companies should operate in order to avoid violating copyright and trademark law, and represents a major victory for artists in asserting their rights in their original creations and the potential value derived therefrom.

Given the rapid pace of technological advancement and expanding use of AI, new questions and controversies will undoubtedly arise, however, particularly as AI plays a more significant role in the content creation process.

The plaintiffs in this case are represented by Joseph Saveri and Matthew Butterick of the Joseph Saveri Law Firm. The defendants are represented by attorneys from McDermott Will & Emery; Durie Tangri; Fried, Frank, Harris, Shriver & Jacobson; Ropes & Gray; Cooley; and Latham & Watkins.