Getty sued Stability AI for copyright in the UK — training happened elsewhere

Getty Images sued Stability AI in the UK in 2023, arguing its Stable Diffusion image generator was trained on millions of Getty's copyrighted photos without permission, and that its outputs could reproduce Getty's watermarks. Getty framed it as a landmark test of whether AI training on scraped images infringes copyright.

The claim depended on the training happening on UK soil — and it hadn't. Stability AI's training ran outside Britain, so Getty had to abandon its primary copyright infringement claim mid-trial for lack of territorial jurisdiction. The High Court also found that Stable Diffusion's model weights are statistically trained parameters, not stored copies of the photographs, undercutting the secondary claim too.

On November 4, 2025, the court ruled almost entirely for Stability AI: copyright claims failed, and only a narrow trademark finding survived — some early Stable Diffusion versions could generate Getty/iStock watermarks — but even that carried no damages or injunction. Two years of litigation over the central AI-training question ended without answering it.