Getty Images, a distributor of stock images, filed a complaint against Stability AI Inc. on 6th Feb 2023, accusing it of exploiting more than 12 million Getty images to train its Stable Diffusion AI image-generation system. This is the second case after Getty images filed a lawsuit against Stability AI for misusing images to train AI that is submitted to the U.S. District Court for the District of Delaware.
Stable Diffusion is an open-source neural network created by Stability AI that can produce visuals in response to text cues. The neural network used by Stability AI was trained on many billion photographs, including images that were purportedly taken from Getty Images unauthorized. The UK-based business also debuted DreamStudio, an image creator, in August 2022.
As per the lawsuit, Stability AI, as part of its efforts to establish a competing product, has copied more than 12 million images from their collection along with related captions and metadata, without the company’s consent or payment to Getty. The Getty Images watermark, which can be seen on each image in the latter company’s visual asset catalog, is also allegedly infringed by Stability AI, according to the lawsuit. Stable Diffusion’s output includes a modified version of a Getty Images watermark, which makes it unclear where the images came from and erroneously implies that it is affiliated with Getty Images.
Getty has supposedly asked the court for an injunction to stop Stability AI from using the images and watermarks, in any way, in addition to monetary damages. Additionally, Getty has also asked the court to issue a directive ordering Stability AI to destroy all iterations of its Stable Diffusion neural network that were improperly trained using Getty Images pictures.
Also read: Microsoft and OpenAI officially extend partnership
As Stability AI faces a lawsuit by Getty images for misusing images to train AI, Midjourney Inc. and DeviantArt Inc., who offer AI-powered picture-generating services, are also being sued for copyright in California. The corporations allegedly trained their individual AI models using billions of copyright photographs without getting the required permissions.