Getty Images v Stability AI – The Trial is Over, But What is its Relevance for Copyright Law in the UK and Internationally?

The date of June 9, 2025, was bookmarked in the diary of many intellectual property lawyers worldwide as it marked the start of the long-awaited trial in the case of Getty Images (US), Inc. and Others v Stability Al Ltd. The trial took place over three weeks before Mrs Justice Joanna Smith, raising pertinent issues regarding the intersection of artificial intelligence (AI) technology and copyright law. It has highlighted critical points of contention relating to the training of generative AI models without the permission of the owners of the IP in the training data and how existing UK copyright laws apply to modern generative AI systems. The case also addresses other IP issues, such as trademark infringement, passing off, and database rights infringement. However, this article focuses solely on the copyright claims.

Copyright law is to reward authors for the creation of original works when made, by granting them (or their assignees) exclusive rights to control how their works are made available to the public. Policymakers acknowledge the importance of this. However, policymakers also recognise the importance of supporting the development of world-leading AI models, which require wide access to high-quality data. This has put policymakers in the UK and globally at a crossroads on how best to achieve both objectives lawfully. Consequently, the outcome of this trial could play a crucial role in shaping how the UK Government resolves its AI and copyright conundrum and could influence the international perspective.

Below is an overview of the copyright issues in the case, how they played out at trial, and potential implications for tech companies and copyright holders worldwide.

Who is Stability AI and what is Stable Diffusion?

Stability AI is a UK-based artificial intelligence company that develops Stable Diffusion AI models, which are open-source image generation tools capable of creating or altering images based on text or image prompts. Stability AI trained the original Stable Diffusion model on a dataset containing billions of images scraped from the internet.

Getty’s copyright claims

Prior to the commencement of the trial, Getty’s copyright claims were:

• Primary infringement by Stability AI (i) copying during the training and development stage of Stable Diffusion and (ii) authorising users to copy or communicate its works to the public at the output stage.
• Secondary infringement by Stability AI due to the importation, possession and/or distribution of the trained Stable Diffusion model (an allegedly infringing “article”) in the UK.

As the trial progressed, the two primary infringement (and database right infringement) claims were dropped, leaving only the secondary infringement (and trade mark infringement) claims.

What have we learned?

1. Location of training and development is key

Stability AI had admitted that at least some Getty images were used to train Stable Diffusion. However, Getty’s difficulty with its training allegation was proving that an infringing act had occurred in the UK, i.e., whether Getty’s works were copied in the UK during training by an act such as by downloading them onto hardware in the UK. Stability AI argued that no infringing acts took place in the UK in training Stable Diffusion because Stable Diffusion was wholly developed and trained outside of the UK and, as such, there was no infringement of UK copyright in this respect. Towards the end of the trial, Getty withdrew this claim; therefore, it is unlikely we will get a judgment on that issue, but we may get some obiter views. Accordingly, it would seem to follow that the decision of where to train an AI model is of real importance to the question whether copyright in the training data has been infringed. Copyright protection is territorial, and broadly speaking, the jurisdiction where any infringing acts take place will affect the risk of copyright infringement, and in particular the extent to which exclusions to copyright and defences to copyright infringement are available.

Internationally, there are differing approaches to the exemptions to copyright protection for text and data mining (TDM) purposes. Reproducing copyright works during AI training may be exempted in countries with more expansive exceptions (e.g. the EU’s commercial use TDM exception assuming no valid opt-out has been effected, and the Fair Use doctrine in the US), but it will only be possible to take advantage of such exceptions if all acts of reproduction are restricted to that jurisdiction which means that nothing can be downloaded or stored elsewhere (e.g. in the cloud where it may be unclear where the data is stored). Moreover, these TDM exemptions will only apply (if at all) to the steps of training the AI model. Subsequent acts of reproduction or communication to the public following completion of the AI training may involve new infringing acts that may not be covered by the TDM exceptions.

2. Technical solutions could mitigate the risk of an injunction

Getty dropped the output claim supposedly because Stability AI has now successfully blocked all use of the disputed prompts on its current cloud-based versions of its model, making an injunction unnecessary. Implementing technical solutions retrospectively could help mitigate the risk of an injunction and consequently the impact on the business’s revenue and reputation.

3. Check your chain of title before issuing proceedings

A big part of the trial, including seventy pages of Getty’s skeleton, was devoted to the chain of title of the copyright in the works which Getty claimed had been copied. As an international group with various subsidiaries, proving chain of title can be complex, as evidenced by the fact that by the end of the trail, title to the copyright in nine of the thirteen infringing works alleged to be infringed remained in dispute.

Therefore, let this be a lesson to all companies, in particular global companies with multiple subsidiaries, to ensure they have resolved any issues regarding ownership of the copyright in the works before issuing proceedings. Otherwise, you might find that the party bringing the action does not have the right to bring it in the first place.

4. Deployment into another territory may amount to copyright infringement

Due to Getty withdrawing its training claim, the core of the UK Getty case is now whether UK copyright law bites on the importation and use of an AI model that was trained elsewhere but which has been made available to UK consumers. This question is critical for AI developers as even if most AI training occurs in the U.S. or China (as it currently does), the UK still remains a valuable commercial market and the court’s interpretation on this issue will no doubt influence policymakers elsewhere.

To prove secondary infringement, it is fundamental to show that Stability AI knew or had reason to believe that Stable Diffusion comprises an infringing copy of Getty’s works. Therefore, the key questions for the court to grapple with are:

• Whether Stable Diffusion is an “article” under English copyright law for the purposes of secondary copyright infringement; and if so,
• Is it sufficient that it is common knowledge that Getty’s images are protected by copyright, and that copyright in them would be infringed by copying them, or by the importation of infringing copies of them?

How the court rules on these questions will clarify whether AI models trained in other countries but intended to be made available to UK users will infringe under UK copyright law, and if so, how.

Concluding thoughts

AI developers should conduct due diligence and seek legal advice to ensure compliance with copyright laws in the countries where their AI models are trained and developed. For example, international companies should ensure that employees are not downloading and storing such data in the UK, including on their mobile devices.

Even if training and development occur outside a territory like the UK that may prohibit such acts (because, for example, there is no applicable TDM exception), the English court may nevertheless rule in favour of Getty regarding its secondary infringement claim (i.e., that an intangible thing such as an AI model can be an “infringing article” and that its deployment into the UK constitutes secondary infringement). This risk affects not only AI developers but also entities which import AI models and related content into the UK. That said, UK policymakers may decide to intervene by passing legislation to address this risk before we receive the judgment, which is due to be delivered in the next court term, although this seems unlikely at the time of writing.