Getty Images v Stability AI [2025] EWHC 2863 (Ch), was a much anticipated judgment, with the decision handed down on 4 November 2025. The case, brought by Getty Images against Stability AI, was broadly about Getty’s concern in protecting its intellectual property when allegedly used by Stability AI for the purpose of training artificial intelligence systems.

The initial proceedings concerned “primary and secondary copyright infringement, database right infringement, trade mark infringement and passing off” but many were eventually discontinued by Getty. By the trial in June 2025, one of the first against an AI developer to be heard, only two of the claims remained.

1. Secondary infringement of copyright 
2. Trademark infringement dues to ‘watermark-like features’.

Getty is a global content creation brand that includes iStock which broadly speaking licences rights to photos and videos. Stability AI is a leading AI company, with various products including Stable Diffusion. Per the judgment, Stable Diffusion is a “type of generative AI model known as a diffusion model, or more specifically a latent diffusion model” which creates outputs (the AI generated image or video) by using real images and videos as samples, from which it then generates an AI version.

This “typically involves designing and building the architecture for the model which is then trained by repeated exposure to massive quantities of data, in this case in the form of human-generated digital images contained in datasets created by crawling and scraping images and associated descriptive captions from the Internet”.

Getty’s claim suggested that their images and videos were included in this database. However, Stability AI were able to advance the argument that any development and training of the database took place outside the UK, and therefore UK copyright and database rights were not infringed.

With the case narrowed dramatically, Mrs Justice Joanna Smith did not find in Getty’s favour in the matter of the secondary infringement, because Stable Diffusion did not itself store or reproduce Getty’s works and therefore was not an “infringing copy”. Getty succeeded in a limited fashion in relation to trademark infringement. She called her findings both “historic” and “limited in scope”.

However, the case leaves unresolved the broader question of whether training generative AI models on copyrighted works within the UK constitutes infringement. Clarity is still required, and does not mean that AI developers are able to use copyrighted datasets. The government has separately said that “Uncertainty over how our copyright framework operates is holding back growth for our AI and creative industries. That cannot continue.” The results of the earlier consultation, which closed in Feb this year, are due in 2026.