The ruling in Getty Images v Stability AI is seen as a setback for copyright holders as calls grow for new UK rules on AI training data
A London-based artificial intelligence company has won a high-profile High Court case examining whether AI developers can lawfully train their models using vast libraries of copyrighted material.
Stability AI, whose board includes Avatar filmmaker James Cameron, successfully defended a lawsuit brought by Getty Images that claimed the company violated copyrights by cracking millions of its photos to train the Stable Diffusion image-generation model.
Ms Justice Joanna Smith found that Getty had failed to demonstrate that the training had taken place in the UK and ruled that the resulting AI model did not constitute an “infringing copy” under applicable law. However, Getty found success with limited trademark claims after it was discovered that some AI-generated images contained replicas of the Getty watermark.
The ruling is seen as a blow to copyright holders’ ability to control how their work is used in AI training. Rebecca Newman, legal director at Addleshaw Goddard, said it underlined that “the UK’s secondary copyright regime is not strong enough to protect its creators.”
Evidence presented in court showed that Getty’s images were used to train the Stability model, which creates images from text prompts. Getty argued that Stability was “completely indifferent” to what it was recording, but the judge said the dispute underscored a broader societal question about “where the balance should be struck between the creative industries and the AI sector.”
The decision comes amid intense debate over how the Labor government should legislate to meet the competing interests of artists and AI developers. Figures such as Elton John, Kate Bush, Dua Lipa and Kazuo Ishiguro have called on ministers to protect creative rights, while tech companies have argued for widespread access to copyrighted data to develop powerful generative models.
The government is currently consulting on proposals to create an exception for “text and data mining” in UK law, which would allow copyrighted works to be used for AI training unless rights holders specifically object.
In a statement, Getty said it remains “deeply concerned” about the lack of transparency rules around the use of AI data.
“We have invested millions of pounds to get to this point, with only one provider to pursue in another location,” the company said. “We call on governments, including the UK, to introduce stricter transparency requirements to prevent costly litigation and enable creators to protect their rights.”
Christian Dowell, Stability AI’s general counsel, welcomed the decision: “Getty’s voluntary dismissal of most of its copyright claims left only a subset of the questions before the court, and this final decision resolves the copyright concerns that were at the heart of the case.”
Gosia Evans, senior lawyer in Harper James’ intellectual property team, said the ruling still leaves the crucial question unanswered: “Should training AI models for copyrighted works be lawful in the UK? We need a government that regulates AI use in a practical and forward-looking way, without removing protections that are vital to the UK’s creative industries.”
The case is likely to influence future disputes over how AI systems obtain their data – and increase pressure on policymakers to clarify where copyright ends and machine learning begins.
