Introduction to the Getty Images AI Copyright Case
When the High Court opens its doors to Getty Images and Stability AI this Monday, it won’t just be hearing a copyright claim. It will be interpreting how UK intellectual property law applies to a technology few legislators could have imagined when the law was written: generative artificial intelligence, especially in context with developments expected in June and the order made in April.
At the heart of Getty’s lawsuit is the claim that Stability AI infringed its intellectual property rights and copyright works by using millions of Getty images and captions — allegedly scraped from its website — to train its text-to-image model, Stable Diffusion. Getty alleges approximately 7.5 million visual assets were lifted, forming part of the model’s training data, without license or permission.
UK Copyright Law and the Issue of Training Governance
UK copyright law protects original artistic works, including photographs, under the Copyright, Designs and Patents Act 1988 (CDPA). The core question before the court is whether copying protected works — even if done algorithmically and not for direct human viewing — for the purpose of training an AI model constitutes infringement under the CDPA.
Under Section 16 of the Act, copyright holders have the exclusive right to copy, issue, or adapt their work. Getty’s argument is simple: training an AI model necessarily involves making copies of the data, even if only temporarily or as part of a dataset. And if that data contains copyrighted images, then the copying is an infringement unless covered by an exception or a license.
This puts the focus squarely on Section 29A CDPA, which covers the exception for “text and data mining” (TDM) for non-commercial research. That exception doesn’t apply here — not just because Stability is a commercial entity, but because Getty argues the act of training the AI was for direct commercial use: the development of a product.
The “Substantial Part” Test and AI Outputs
But the claim isn’t just about inputs. Getty says that Stable Diffusion has also produced images that incorporate “substantial parts” of Getty’s original works, including occasionally reproducing Getty’s watermark — the clearest sign, in their view, that the model was trained on protected material without consent.
Under UK law, it’s not necessary for a whole work to be copied to constitute infringement. The test is qualitative: has a substantial part of the expression, not the idea, been reused? That’s a high bar to meet in the context of AI-generated content, but if any of the 17 images Getty is focusing on resemble the originals too closely, it could sway the court.
Jurisdiction and Extraterritorial Reach
Stability is challenging the claim in part by asserting that its model was trained as part of stability in the United States, not the UK. Under UK conflict of law principles, this raises a significant jurisdictional question: can a UK court rule on acts of infringement that occurred abroad?
Typically, UK copyright law only applies to infringing acts that occur within the United Kingdom and, in some cases, France. Getty must convince the court that there is a sufficient connection to the United Kingdom — possibly through the availability of Stable Diffusion or its outputs within the UK market — to establish jurisdiction. It’s a test that may reshape how far UK IP law reaches in a borderless digital ecosystem.
Getty’s Claims and Stability AI’s Defence
Getty’s claims centre on the assertion that Stable Diffusion not only leveraged its images without permission but also produced derivative works that mirror the original content too closely to qualify as fair use. The gallery emphasises the notion of “substantial parts,” suggesting that even fragments of their imagery—especially those containing watermarks—demonstrate direct infringement. On the other hand, Stability AI is poised to mount a vigorous defence, arguing that its model operates under principles of transformative use.
Copyright, Database Rights, Trade Marks, and Passing Off
In the unfolding saga of Getty Images vs. Stability AI, the stakes hinge on complex copyright laws and digital ownership. Getty’s accusations focus not just on theft but on the very essence of creativity, arguing that Stability’s outputs are not mere reproductions but derivative works that infringe upon their intellectual property. By highlighting the presence of watermarks within such images, Getty aims to showcase a direct line from its original creations to the alleged infringements. Conversely, Stability AI contends that its model embodies transformative use—a doctrine allowing for certain adaptations under specific conditions.
Limits of Collective Action and Late-stage Evidence
Getty originally sought to include a class of up to 50,000 photographers whose work had allegedly been used. The court denied this, narrowing the claim to focus only on Getty’s own rights. In January, Judge Joanna Smith also blocked the company’s attempt to expand its evidence base late in the proceedings, pushing back against any attempts to broaden the scope beyond what had already been agreed.
That procedural discipline underscores the court’s focus: this is a case about direct copyright infringement, not a general inquiry into AI ethics or artistic fairness.
The Legislative Undercurrent
While the High Court deliberates on what the law is , Parliament is already grappling with what the law should be . The House of Lords has repeatedly attempted to introduce transparency obligations into the Data Protection and Digital Information Bill, aimed at forcing AI developers to disclose whether they have used copyrighted materials in training.
These amendments — driven by concerns from creatives and spearheaded by filmmaker and peer Beeban Kidron — reflect a growing sense in the UK creative industries that their work is being co-opted by powerful tech companies without compensation or consent.
The Case’s Impact on Businesses and Content Creators
The outcome of the Getty Images case could set a critical precedent for how businesses and content creators engage with AI technologies. If the court favours Getty, it may strengthen copyright protections, compelling AI developers to seek explicit permission from creators before utilizing their works for training purposes. This could lead to more stringent licensing agreements and a potential shift in how AI models are built. Conversely, should Stability AI prevail, it might embolden tech companies to interpret transformative use broadly, allowing them to freely incorporate existing works into their systems without fear of litigation.
Innovation v. Protection: A Legal Crossroads
What happens in court over the coming days may help to clarify whether training AI on copyrighted works without permission is simply part of technological progress — or a breach of fundamental legal protections.
As Gill Dennis of Pinsent Masons LLP puts it, “the court’s decision will test the jurisdictional reach of UK copyright law and provide clarity,” sending signals not just to developers, but to lawmakers considering how to support both innovation and IP protection.
Whether the court rules in favour of Getty or Stability AI, the case marks a pivotal moment. It forces the legal system to confront the question many have been asking: who owns the future of creativity — the humans who made the content, or the machines learning from it?
Practical Approaches to AI Compliance
In the face of this impending legal battle, content creators and AI developers alike must consider practical approaches to ensure compliance with copyright regulations. Both parties can benefit from establishing transparent communication channels that facilitate a mutual understanding of rights and usage. Content creators should have robust licensing agreements that clearly outline how their works may be used, whether for training AI models or other purposes. On the other hand, AI companies could adopt ethical guidelines that prioritise obtaining consent from creators before utilising their intellectual property. This proactive strategy not only fosters goodwill but also helps mitigate potential legal repercussions stemming from unauthorised use of copyrighted material.
What are the main legal arguments in the Getty Images AI copyright case?
The primary legal arguments in the Getty Images AI copyright case centre on the unauthorised use of copyrighted images to train AI models. Plaintiffs argue that this infringes on intellectual property rights, while defendants claim fair use protections apply. The outcome could significantly impact AI development and copyright law interpretations.
