Getty Images has recently pared down its legal challenge against Stability AI at London’s High Court, marking a shift in its strategy over the use of copyrighted images in training AI models like Stable Diffusion. By dropping several claims, Getty acknowledges the difficulty in proving that AI-generated outputs replicate key elements of original works — a critical point when determining copyright infringement.
Yet the fight isn’t over. Getty continues to pursue secondary and trademark infringement claims both in the UK and through a separate lawsuit in the US. The argument is that by incorporating millions of images into its AI, Stability AI might be importing infringing material into the UK, regardless of where the training took place. Across the Atlantic, Getty seeks roughly $1.7 billion in damages over allegations that its content was used without permission to train the AI.
Stability AI has taken a measured stance about Getty’s move. The company is optimistic that the remaining claims, particularly those involving trademarks, will not hold water given that users don’t generally equate watermarking with commercial endorsements from Stability AI.
This case comes on the heels of a US court ruling that favoured Anthropic in a similar dispute involving unauthorised AI training on copyrighted books. For anyone who’s ever faced uncertainty around digital rights and copyright, this evolving battle is a reminder that the legal framework is still catching up with fast-changing technology.
Getty, meanwhile, is adapting to the new landscape by offering its own AI-powered tool that generates licensable images from its iStock collections — a move that highlights the industry’s shift towards integrating advanced tech solutions while navigating copyright complexities.
