UK Ruling in Getty Images' AI Lawsuit Leaves Key IP Questions Unanswered

A recent UK High Court ruling in the case of Getty Images v. Stability AI has concluded with a mixed verdict, leaving several critical questions about intellectual property (IP) rights in the age of artificial intelligence unanswered. The case, filed in 2023, revolved around Getty Images' claims that Stability AI, the company behind the Stable Diffusion image generation model, had infringed upon its copyright and trademark by using Getty's images to train its AI and by generating outputs that included Getty's watermarks.

While the court acknowledged some trademark infringement, it largely dismissed Getty's copyright infringement claims, a decision that has sparked debate among legal experts regarding the scope of copyright protection in the context of AI. Justice Joanna Smith of the UK High Court of Justice described the case as "historic" due to the novel issues raised by AI. However, she underscored that her "findings at this trial are, of necessity, extremely limited" because Getty Images only offered evidence of alleged infringement it generated through its investigation (and not by actual users in real life) and only in respect of the earlier v1.x and v2.x of Stability AI models.

Getty had argued that Stability AI's Stable Diffusion system was trained using images scraped from Getty's website, and that the AI-generated images sometimes reproduced Getty's copyrighted images and watermarks. However, Getty dropped its primary copyright claim mid-trial due to a lack of evidence proving where the training of Stable Diffusion had taken place. This led to the dismissal of the more general claim of copyright infringement. "Getty Images may be able to maintain such a case in the jurisdiction where the Model was in fact trained, but there is no basis for that case in this jurisdiction," Smith wrote in her decision.

The remaining claim of secondary copyright infringement, which alleged that Stability AI had imported an AI model into the UK that breached Getty's copyright, was also dismissed. The court reasoned that Stable Diffusion does not store or reproduce copyrighted works, instead recording the weights, and therefore does not constitute an infringing copy under the UK's Copyright, Designs and Patents Act (CDPA) of 1988. "An AI model such as Stable Diffusion which does not store or reproduce any copyright works (and has never done so) is not an 'infringing copy'," the judge ruled.

Despite the loss on copyright grounds, Getty Images secured a partial win on its trademark infringement claim. The court found that the presence of Getty watermarks on certain images generated by Stable Diffusion models constituted trademark infringement, affirming that responsibility for the presence of such trademarks lies with the model provider. However, the judge noted that these findings were "extremely limited in scope". The Justice found too speculative the possibility that Stability AI's model would produce in real images of Miley Cyrus that were NSFW (not safe for work) that included also Getty watermarks. Getty was able to produce only 19 such examples, but there was no evidence of such images being generated by users in real life.

The ruling has been met with mixed reactions. Some legal experts believe it weakens intellectual property laws, while others see potential for trademark and copyright protection in the judge's ruling. Intellectual property experts said the decision clarified, at least for now, how UK law treats the use of copyrighted material in AI training. Simon Barker, Partner and Head of Intellectual Property at law firm Freeths, described the case as "where intellectual property meets AI". He said the High Court had "drawn a line" by confirming that training an AI model on copyrighted works does not constitute secondary infringement, provided the works are not stored or reproduced in the model itself.

Getty Images expressed its concern that even well-resourced companies face significant challenges in protecting their creative works due to the lack of transparency requirements. The company urged governments to establish stronger transparency rules that would require AI developers to disclose the data sources used in training their models.

Stability AI welcomed the decision, with its general counsel stating that the ruling "ultimately resolves the copyright concerns that were the core issue".

The case highlights the complex legal challenges posed by generative AI and the need for clearer legal frameworks to address copyright and trademark issues in this rapidly evolving field. While the UK ruling provides some guidance, it leaves significant questions about AI training and intellectual property open for further discussion and potential future litigation.