Getty Images drops its primary claims of copyright infringement against Stability AI in UK court but is still pursuing other claims and a separate US lawsuit

On its first morning of closing submissions, Getty has dropped its primary copyright infringement claims against Stability AI. … See also Mediagazer

TechCrunch 2025-06-26 Rebecca Bellan

Context & Ripple Effects

Getty’s UK action followed its 2023 plan to challenge Stable Diffusion’s creator in Britain and a separate US complaint over alleged training-data copying. The London case had only recently entered trial, with Getty alleging that millions of images were scraped and Stability AI denying the allegations.

The withdrawal narrows one major part of a long-running dispute rather than ending it: Getty retains other UK claims and its separate US case. That makes the precise legal basis and venue of AI-training challenges central to the dispute’s next phase.

First-order effects

Getty will no longer seek a ruling on its primary UK copyright-infringement claims, reducing the issues the High Court must resolve in this proceeding.
Stability AI still faces Getty’s remaining UK allegations and the separate US litigation, so the withdrawal does not remove the underlying legal exposure.

Second-order effects

The narrowed UK case gives both parties a more limited near-term test of claims arising from image-model development, while the US action remains a separate route for Getty’s broader allegations.
Other image-model developers and rightsholders will watch which remaining claims survive, since the case’s practical value may shift from a single sweeping copyright theory to claim- and jurisdiction-specific arguments.

Third-order effects

If AI-training disputes continue to fragment across legal theories and countries, licensing, model-development, and litigation strategies are likely to be shaped by venue-specific risk rather than one universal copyright answer.
The episode underscores a durable uncertainty in the market: courts may define the boundaries of training-data use incrementally, leaving commercial agreements to manage risk where litigation produces only partial clarity.

The trend: Generative-AI copyright disputes are moving from broad challenges to more granular tests of training-data claims across jurisdictions.

Discussion

@technollama Andres Guadamuz on x
I'd also wager that this case could have an effect on the ongoing copyright consultation. No AI company will perform any training in the UK after this result, so whatever is decided in the consultation will be moot as even any exception would have practically no effect.
@technollama Andres Guadamuz on x
This is big because it gives us an indication of the problems that future copyright infringement cases will have in the UK. Getty have dropped the case because they weren't able to prove that training took place in the UK, so no jurisdiction for copyright infringement.
@shakeelhashim Shakeel on x
Pretty terrible week for AI copyright advocates
@technollama Andres Guadamuz on x
Huge news coming out of the Getty v Stability AI trial in the UK. Getty have dropped their primary copyright infringement claims in their trial, and will be relying on the trade mark claims. https://www.linkedin.com/...
@technollama Andres Guadamuz on x
Whenever we illustrate the Getty v Stability case in the future, this is the image that will be used, which tells you everything you need to know. I won't even mention that it features a Tottenham player, which is unforgivable in its own right. [image]
@dorialexander Alexander Doria on x
Looks to be clarification season for AI training and copyright: *Fair use/Google books doctrine holds on in the US. *In all likelyhood, text & data mining exception will apply to AI training [EU] *Jurisdiction only applies when training take place on location [UK]

Chronicles