A group of artists has initiated a class action lawsuit against Google, claiming that the tech giant's Imagen AI model was developed using their copyrighted works without obtaining proper permission. This lawsuit, filed in a California District Court, argues that the text-to-image model constitutes an “infringing derivative work,” as it allegedly includes reproductions of their images that were duplicated numerous times without consent.
The artists contend that Google has engaged in “massive copyright infringement” through its Imagen model and are seeking damages for what they consider the company's “wrongful conduct.” The lawsuit asserts that the images in question were utilized not only for training Imagen but also for other AI systems, including Imagen 2 and the multimodal foundation model, Gemini.
The complaint emphasizes that the plaintiffs never authorized Google to utilize their copyrighted artwork. "Google repeatedly violated plaintiffs’ exclusive rights… and continues to do so today," it states.
Launched in May 2022, Imagen was introduced amidst the rising popularity of text-to-image models like Stable Diffusion and Midjourney, with Imagen 2 debuting the following December. This image generation model is integrated into Google Cloud platforms such as Vertex, which allows users to create images for various business applications. Additionally, Imagen serves as the foundational technology for generative 3D models like Google DreamFusion.
The dataset LAION-400M, an open-source collection of URLs to images for training models, was reportedly used in developing Imagen. The artists assert that their works were part of this dataset, and by copying LAION-400M to construct its own AI systems, Google effectively created unauthorized copies of their art.
According to the lawsuit, “The intermediate copies of each copyrighted work that Google made during training of the Google LAION Models were substantially similar to that copyrighted work.” Consequently, the plaintiffs claim to have suffered injuries resulting from Google's actions.
As part of their demands, the artists are requesting that Google eliminate all reproductions of their works and cover their legal expenses. They are represented by attorneys who previously advocated in the 2022 GitHub Copilot lawsuit, as well as legal actions against OpenAI, Meta, Nvidia, and Databricks on behalf of authors in the publishing space.
In addition to targeting Google, Joseph Saveri and Matthew Butterick have also filed copyright infringement lawsuits against other image generation platforms, including Midjourney, Runway ML, and Stable Diffusion, representing the rights of artists in the evolving digital landscape.