Getty Images Takes On Stable Diffusion For Copying 12M Photos, Famed Watermark

Several weeks ago, Getty Images filed a lawsuit against Stability AI, the maker of artificially-intelligent art generator Stable Diffusion, alleging that the algorithm had illegally scraped its extensive library to train the latter’s database. 

At the time, the company said it suspected the art generator “unlawfully copied and processed millions of images protected by copyright.” The latest dispute claims that number reached a whopping 12 million photographs. 

More importantly, the complaint states that aside from training its software on the pictures without consent, Stability AI might have stolen associated captions and metadata “as part of its efforts to build a competing business.” 

According to Ars Technica, documents filed assert that the firm might have gone as far as removing Getty’s copyright information, falsifying its own claims, and infringing upon the “famous trademarks” by duplicating the watermark on images. 

The publication notes that by copying the detailed descriptions and metadata from photographs, Stability AI could use this information to better respond to prompts in the future, making itself a bigger competitor as an image provider. 

Despite it taking on the image generator in these instances, Getty Images previously emphasized it was not against AI as a whole and was convinced it “has the potential to stimulate creative endeavors.” 

Instead, it believes Stability AI, while positioning itself as a competitor for stock images, has used the technology in the wrong manner, and did not pay for the rightful licenses to train the algorithm on its extensive collection.  

[via Ars Technica and Reuters, cover image via Eric Broder Van Dyke | Dreamstime.com]