Getty Images has largely lost its legal battle against Stability AI, the company behind the popular image generator Stable Diffusion. While Getty scored a narrow win on trademark grounds, the broader copyright claims that had many creators on edge were dismissed, leaving both sides claiming partial victory and lawyers calling for urgent reform of UK copyright law.
At the heart of the case was a fundamental question: Can an AI company legally use copyrighted images to “train” its models?
Seattle-based Getty Images, known for its vast library of editorial and creative content, accused Stability AI of using millions of its licensed photographs without permission to train Stable Diffusion, a model that creates images from text prompts.
Getty argued that this amounted to copyright infringement, since the AI system allegedly “learned” from their photos and sometimes reproduced identifiable content.
But during the trial, Getty dropped this core argument. It couldn’t prove where Stability AI had trained its system or whether Getty’s images were directly involved. Without that evidence, the case’s most pivotal question, whether training AI on copyrighted data is itself illegal, was left unanswered.
Two intellectual property lawyers quoted by Reuters described the result as “a damp squib” for copyright law, meaning the long-anticipated test case failed to produce the clear precedent many had hoped for.
Judge Joanna Smith ultimately ruled that Getty had succeeded “in part”, but only on a narrow trademark issue.
Her judgment found that AI-generated images showing fake Getty watermarks amounted to trademark infringement. However, she was quick to stress that these findings were “historic and extremely limited in scope.”
More significantly, she dismissed Getty’s broader claim that Stable Diffusion itself infringed copyright by reproducing works. Smith concluded that “Stable Diffusion… does not store or reproduce any copyright works.”
That line, simple but powerful, exposes a key challenge in applying traditional copyright rules to modern AI systems. As lawyer Rebecca Newman of Addleshaw Goddard quoted by Reuters put it, “Today’s finding means that copyright owners’ exclusive right to reap what they have sown has been avoided on a technicality.”
In other words, because AI models don’t keep literal copies of images, they may slip through the legal cracks, even if they were trained on copyrighted material.
Mixed reactions: “A win,” but for whom?
Both sides, predictably, claimed a form of victory.
Getty Images said the ruling confirmed that Stability AI’s model had used its trademarks, a “significant win for intellectual property owners” in their view. The company also noted that the decision establishes “a powerful precedent that intangible articles, such as AI models, are subject to copyright infringement claims.”
Getty plans to use that argument in its parallel lawsuit against Stability AI in the United States, where the company hopes for a broader interpretation of copyright protections.
Stability AI, meanwhile, also welcomed the outcome. Its general counsel, Christian Dowell, said the judgment “ultimately resolves the copyright concerns that were the core issue.”
But the markets seemed to take a different view, Getty’s shares fell nearly 9% in premarket trading following the news, suggesting investors see the ruling as a setback for the company’s broader legal campaign.
This case is just one of several ongoing battles over the use of copyrighted content in training generative AI. Similar lawsuits are active in the U.S. and Europe, where artists, writers, and photographers are pressing for compensation and control over how their work is used.
The UK ruling, though limited, points to a major gap in current law: it doesn’t clearly say whether training an AI model on copyrighted material counts as infringement.
In short, there’s still no definitive answer to the question everyone is asking: Is teaching an AI system with copyrighted data legal?
Getty’s statement after the ruling struck a note of concern about the difficulty of enforcing rights in the digital age. “Even well-resourced companies such as Getty Images face significant challenges in protecting their creative works,” the company said.
It also urged governments, including the UK, to require more transparency from AI developers, particularly around what data is used to train their systems. Without such rules, Getty warned, creators could face “costly legal battles” just to find out if their work was used.
This reflects a broader industry debate: AI systems are only as ethical as the data they’re trained on. Yet, few companies disclose their training datasets, citing proprietary concerns or technical limitations.
