AI sued by image library for intellectual property infringement in training models
7 November 2025
Stability AI (Defendant) [2025] EWHC 2863 (Ch)
The legal dispute between Getty Images (and its associated companies) and Stability AI revolves around complex issues of copyright infringement, database rights, trademark infringement, and passing off. The arguments centred on the use of Getty Images’ visual content in the training and operation of Stability AI’s generative AI model, Stable Diffusion. Media firm Mischcon de Reya has acclaimed this as the “one of the most anticipated cases in recent years.” The case has significant implications for intellectual property law as it intersects with the development and deployment of AI technologies in the UK.
Background and Parties
The claimants in the case are several related companies under the Getty Images brand. These entities collectively own or have exclusive licenses over millions of high-quality photographic and artistic images referred to as the “Visual Assets” or “Copyright Works.”
Stability AI Limited, the defendant, is a UK-based company that developed the Stable Diffusion AI model, which is a deep learning image generation tool that creates images based on text or image prompts, including around 12.3 million visual assets, together with associated captions, from the Getty Images websites, as well as publicly accessible third-party websites.
According to Getty Images Stability AI scraped millions of their copyright-protected images from its websites without authorisation.
The Core Claims
Getty Images initially brought a broad claim including allegations of primary and secondary copyright infringement, database right infringement, trademark infringement, and passing off. They argued that:
• Stability AI unlawfully used Getty’s copyrighted works without permission to train the AI model.
• The AI model outputs sometimes reproduced Getty’s images or bore their trademarks (watermarks), infringing Getty’s rights.
• Stability AI’s making of the model weights available for download constituted secondary copyright infringement. (Model weights are the values that determine how inputs are transformed into outputs in a neural network, reflecting the strength and direction of connections between artificial neurons after training. During training, optimisation procedures adjust these weights so the model improves at a task; the final set of weights effectively encodes the model’s learned “knowledge” from data. These “weights” are machine-readable parameters, distinct from source code text; they are large arrays of numbers that operationalise the model’s behaviour rather than human-authored narrative code.
• Use of Getty’s trademarked watermarks within generated images constituted trademark infringement.
As the judge observed,
Both sides emphasise the significance of this case to the different industries they represent: the creative industry on one side and the AI industry and innovators on the other. Where the balance should be struck between the interests of these opposing factions is of very real societal importance. Getty Images deny that their claim represents a threat to the AI industry or an attempt to curtail the development and use of AI models such as Stable Diffusion. However, their case remains that if creative industries are exploited by innovators such as Stability without regard to the efforts and intellectual property rights of creators, then such exploitation will pose an existential threat to those creative industries for generations to come.” [para 12]
In her summary of the judgment, Nina O’Sullivan of Mischcon de Reya observes that attention will now turn to the response to the government’s consultation on copyright and GenAI, as it faces pressure from creative industries opposing a general text and data mining exception that would allow AI companies to scrape copyright works unless rights holders expressly opt out.” Getty Images v Stability AI: Unpacking the High Court’s judgment
Narrowing of Issues and Trial Developments
During the trial, Getty Images abandoned the claim that the training and development of Stable Diffusion took place in the UK due to no evidence supporting this. They also dropped primary copyright infringement claims concerning the AI-generated outputs. The trial then mainly focused on:
• Secondary copyright infringement linked to Stability AI’s distribution of the AI model.
• Questions of trademark infringement and passing off, particularly related to watermarks visible in some AI-generated images.
Court Findings and Judgment
In November 2025, Mrs Justice Joanna Smith DBE delivered a 205-page judgment on the case with mixed outcomes for both parties:
The court dismissed the core claim of copyright infringement for the use of Getty’s images in training Stable Diffusion, largely due to jurisdictional and evidential issues (no strong evidence that training occurred in the UK). Stability AI prevailed on this issue, since its Diffusion model was trained overseas and its importation into, and distribution within, the UK did not amount to dealing in an article which was an ‘infringing copy’.
Secondary Copyright Infringement: Getty argued there was no requirement under the Copyright, Designs and Patents Act 1988 that an article must continue to retain a copy of the works and it was sufficient to show that the making of the model weights, had it been carried out in the UK, would have constituted an infringement of copyright. The court rejected this argument: an infringing copy had to be a ‘copy’, i.e. a reproduction of the copyright work. The model weights of the various iterations, in the judge’s view, did not store the visual information in the copyright works.
The judge posed the central question: is an AI model which derives or results from a training process involving the exposure of model weights to infringing copies itself an infringing copy? She concluded that it was not. Whilst the model weights were altered during training by exposure to the copyright works, by the end of that process they did not store any of those copyright works. The model weights were “purely the product of the patterns and features which they have learnt over time during the training process”. The claim for secondary copyright infringement therefore failed.
Trademark Infringement: Getty’s trade mark infringement claim concerned the display of watermarks in outputs, which are identical or similar to its trade marks for GETTY IMAGES and iSTOCK. It argued that this amounted to trade mark infringement pursuant to sections 10(1), (2) and (3) of the Trade Marks Act 1994 (TMA).
To succeed in its trade mark infringement claim, Getty had to establish, as a threshold question and on the balance of probabilities, that Stable Diffusion had generated at least one output containing the relevant watermarks in the UK. The court made a number of model/version specific threshold findings, and concluded that there were some “limited” and “historic” instances of trade mark infringement. Stability AI was therefore found liable for limited trademark infringement related to the inclusion of Getty’s watermarks in early versions of Stable Diffusion.
The court decided not to give judgment on the passing off claims, given that this would not add further to the trade mark claims.
The judgment established an important precedent that AI models and intangible articles like model weights can be subject to copyright infringement claims, similar to tangible goods. However, the ruling exposed limitations in UK law regarding protection of rights against modern AI technologies.
Industry and Legal Implications
The case is notable for being one of the first major intellectual property disputes concerning generative AI and web scraping in the UK. Getty Images characterised the ruling as a partial win but expressed concern over challenges faced in protecting creative works due to lack of transparency in AI training processes.
The judgment highlights deficiencies in current copyright protections and calls have been raised for stronger legislative measures to safeguard IP in the era of AI.
The judge’s finding that using copyrighted data for AI training outside the UK may currently avoid UK copyright liability essentially allows AI developers to continue this practice unless legislated otherwise or caught by other jurisdictions’ laws .
It is also significant that the AI model itself, represented by “model weights,” was not found in the judgment to store or reproduce copies of the copyrighted works in a way that legally constitutes an infringing copy. Therefore, training on copyrighted images does not amount to secondary copyright infringement under UK law. This judicial recognition separates the ephemeral use of copyrighted data in machine learning from unlawful copying, enabling AI developers to argue that their training processes fall outside infringement .
The ruling has exposed gaps and limitations in existing copyright laws concerning AI. Even well-resourced rights holders like Getty face challenges in protecting their works when AI technologies operate in opaque ways with vast datasets. The judgment highlights the struggle under current law to control or prevent the use of copyrighted content in AI without robust transparency or opt-in obligations.
The decision also illustrates how fact-sensitive these types of claims are, particularly where each attempt to obtain evidence of infringement will generate a different output, making reproducibility almost impossible.
