The English High Court has delivered its long-awaited judgment in Getty Images v. Stability AI. The case relates to the alleged infringement of intellectual property (IP) rights, in photos owned by Getty Images, Inc (Getty), by Stability AI (Stability) in the training and use of its deep learning AI model, "Stable Diffusion" (the Model). It has been very closely followed across the legal, creative and AI sectors alike, with the decision widely expected to shed light on how UK IP law applies to the rapidly evolving field of generative AI. However, whilst on the surface, the judgment may appear to have provided some clarity on the interplay between AI and IP infringement, there remains significant uncertainty in reality.
Key Takeaways
- Getty succeeded only in part, and narrowly, on its trade mark infringement claims. Findings were confined to specific examples of outputs from early versions of the Model involving "iStock" and "Getty Images" watermarks.
- Getty's secondary copyright infringement claim failed. The court held that the Model’s weights were not an "infringing copy" of Getty's works because the Model did not at any time contain or store a copy of the underlying Getty images.
- The judgment leaves critical questions unanswered relating to the relationship between IP rights and generative AI, particularly whether the use of copyright protected works to train AI models constitutes copyright infringement. It was hoped that these issues would be addressed by the court, but this element of the claim was withdrawn during trial.
- Looking ahead, the UK government's ongoing work with expert groups from both the creative and technology sectors will be closely watched, as it seeks to balance the protection of human creativity with technological innovation.
Background
Getty, a leading global provider of stock photography representing more than 600,000 photographers and artists, brought proceedings in the English courts against Stability, an open-source generative AI company. Getty alleged that Stability had unlawfully scraped over 12 million Getty-owned images, which it described as "the lifeblood" of its business, to train the Model without Getty’s consent.
Stability accepted that "many" copyrighted works had been used to train the Model and admitted, in its statements of case, that "at least some" of those images were from Getty's websites and were used in training earlier versions of the Model. Yet, Stability denied any wrongdoing and maintained that it neither copied nor stored any of Getty's copyrighted works. Instead, the Model learned from them in an abstracted form, extracting patterns rather than reproducing original content.
Getty's original 2023 claim alleged primary copyright infringement, asserting that Stability had copied millions of Getty-owned images both to train and develop its Model (the Training Claim), as well as in the outputs generated by the Model (the Output Claim). However, the Training Claim was abandoned during the trial due to jurisdictional difficulties; specifically, there was insufficient evidence to show that the training or development of the Model had occurred within the United Kingdom. The Output Claim was also withdrawn, as it was demonstrated that Stability had blocked the prompts said to have produced the disputed outputs.
The Court's Key Findings
1. Trade Mark Infringement
Getty argued that the Model generated images which included watermarks bearing the Getty registered trade marks "Getty Images" and "iStock" (the Getty Signs) and that such use constituted trade mark infringement. The court found that, on balance, at least one user of certain earlier versions of the Model would have produced AI-generated outputs containing a Getty Signs watermark. The court emphasised, however, that the evidence of infringement related to a limited number of examples and that the true scale of such outputs could not be determined. The court confirmed partial trade mark infringement, which it described as "both historic and extremely limited."
In making that determination, the court made a number of interesting findings when applying the law to the facts, namely:
- Whether Stability used the Getty Signs in the course of trade: Stability argued that it did not use either of the Getty Signs in the course of trade (a necessary test to satisfy for a finding of trade mark infringement), but instead simply provided the Model and maintained that it was the user who was then in control of the Model's output. Getty asserted to the contrary, claiming that Stability's Model offered the service of generating images and putting those images on to the market, and that, as such, Stability was using the Getty Signs for its own commercial communication. Getty argued that the case involved active behaviour and control on the part of Stability because: (i) Stability is the entity that trained the Model; (ii) Stability is the entity that could have filtered out watermarked images; (iii) Stability makes the Model available to consumers; and (iv) Stability is the entity making the communication that bears the Getty Signs. The court agreed with Getty on this question.
- Whether Stability used the Getty Signs in relation to goods and services: Stability argued that it did not use the Getty Signs in the context of a commercial activity or "in relation to any goods and services" (another necessary test to satisfy for a finding of trade mark infringement). In its finding against Stability, the court stated that, "Essentially [Stability] contends that the appearance of the watermarks on synthetic images is not 'trade mark use' in that it is not liable to affect the functions of the trade mark, in particular its essential function of guaranteeing to consumers the origin of the goods". The court held that the use of the Getty Signs in the Model outputs will be perceived by the average consumer as a commercial communication by Stability, "[t]he [Getty] Signs are affixed to synthetic images generated by customers owing to the functionality of the Model, itself dependant upon its training data (over which Stability has absolute control and/or responsibility). It is in this way that Stability 'offers and puts synthetic images bearing the Signs on the market' and this is Stability's commercial communication to the customer."
- Comparison of goods and services: For the purposes of comparing Stability's acts to the specification of the Getty registered trade marks, the court held that: (i) Stability used the signs in relation to "synthetic image outputs" and "the provision of synthetic image outputs," and (ii) that "synthetic image outputs" are identical to "downloadable digital illustrations and graphics" and "digital media, digital materials, digital content (namely images)" in Class 9, while "the provision of synthetic image outputs" is identical to "digital imaging services" in Class 41.
2. Secondary Copyright Infringement and "Article" Claims
Getty argued that the Model is an "infringing copy" of Getty works on the basis that it had been imported (primarily through downloading) into the United Kingdom. An important point to note is that Getty did not allege that the Model includes or comprises a reproduction of any of the Getty images. Instead, Getty asserted that the Model was an infringing copy because the making of the model weights would have constituted copyright infringement if it had been carried out within the United Kingdom. The court therefore had to answer two important questions: (i) whether the model weights were an "article" under the legislation, and (ii) if so, whether they were an infringing copy. On (i) the Court held that an "article" is capable of being an electronic copy stored in an intangible form. However, on (ii), the Court held that since the model weights never contained or stored a copy of any of the Getty images, they could not constitute an infringing copy. Accordingly, the court rejected Getty's secondary copyright infringement claim.
The Road Ahead
Getty has not indicated whether it intends to appeal the judgment but has stated it would take forward findings of fact in the UK ruling in their ongoing US case. Getty also expressed deep concern that "even well-resourced companies" such as itself "face significant challenges in protecting their creative works," and urged the UK government introduce stronger transparency measures to reduce costly disputes and better safeguard creators' rights1.
Looking ahead, the UK government's ongoing work with industry groups from both the creative and technology sectors will be closely watched, as it seeks to balance the protection of human creativity with technological innovation2. Yet, the debate over IP rights and AI is far from settled, and further developments, whether through appeal, further case law or legislative reform, are inevitable. As ever, it's a matter of watching this space.
*Lavinia Puder, a trainee in our London office, contributed to this advisory.
1 4 November 2025, Press Release, 'Getty Images issues statement on ruling in Stability AI UK litigation'. See here: https://newsroom.gettyimages.com/en/getty-images/getty-images-issues-statement-on-ruling-in-stability-ai-uk-litigation
2 16 July 2025, Press Release, 'Creative and AI sectors kick-off next steps in finding solutions to AI and copyright'. See here: https://www.gov.uk/government/news/creative-and-ai-sectors-kick-off-next-steps-in-finding-solutions-to-ai-and-copyright?utm_medium=email&utm_campaign=govuk-notifications-topic&utm_source=d353b777-fe49-46de-8982-a6fced00770f&utm_content=immediately
London
