Getty Images Files Lawsuit Against Stable Diffusion's AI Art Generator for Alleged Data Scraping Violations

In September, Getty Images prohibited AI-generated works from its commercial database due to copyright concerns. Recently, Getty Images announced a lawsuit against Stability AI, the creator of the popular AI art tool Stable Diffusion, in a London court, alleging copyright violations.

Getty Images asserts that Stability AI unlawfully copied and processed millions of copyrighted images and their associated metadata without a license. This action was taken to benefit Stability AI’s commercial interests, undermining the rights of content creators. In a press statement, Getty Images expressed its belief in the potential of artificial intelligence to enhance creative endeavors but emphasized the importance of respecting intellectual property rights.

The company noted that it has provided licenses to tech innovators for AI training that adheres to legal and ethical standards. However, Stability AI allegedly did not seek a license from Getty Images, opting instead to bypass available licensing options and established legal frameworks.

While the specific details of the lawsuit remain confidential, Getty Images CEO Craig Peters indicated to Tech media outlet that the charges include copyright infringement and violations of site terms of service, such as web scraping. Notably, the company is not pursuing monetary damages; instead, it aims to set a legal precedent for future cases.

Tools like Stable Diffusion, DALL-E, and Midjourney do not create artwork in the same imaginative way humans do; they rely on vast databases of annotated images to generate content. For instance, an algorithm learns what a frog looks like through thousands of labeled frog images. Instead of assembling their own databases, some AI firms take the shortcut of scraping imagery from the internet. Companies like Clearview and Voyager Labs have faced heavy fines for this practice.

An independent study conducted last August revealed that a significant portion of Stable Diffusion's training data likely came from Getty Images, as shown by its tendency to replicate the Getty watermark in generated artworks.