The case revolves around the use of Stable Diffusion, which was allegedly trained on a dataset called LAION-5B, containing over 5 billion images scraped from the web. However, the dataset only contained links and text descriptions, meaning the AI companies would have had to separately gather copies of the images to train Stable Diffusion. The judge denied claims filed under the Digital Millennium Copyright Act of 1998, which prohibits circumventing controls on copyrighted materials. However, the artists argued that the AI models used prompts, including artists’ names, to generate images, potentially constituting copyright infringement.
The ruling has raised questions about the legal implications of AI art generation and the boundaries of copyright law. While it is generally permissible for human artists to create new works in the style of copyrighted artists, the use of AI models trained on specific copyrighted works blurs the line. The case will likely shed light on the training practices and model outputs of AI art generators. Overall, this ruling marks a significant step forward for artists seeking to protect their copyrighted works in the face of AI-generated content.