BREAKING: UK High Court hands down decision in Getty Images v Stability AI

Yesterday, the UK High Court handed down its highly anticipated decision in Getty Images v Stability AI. The headline is that Getty’s claim for secondary copyright infringement failed, but Stability AI was found liable (in part) for trade mark infringement.

In this blog, we take a look at some of the key issues before the court and share our initial thoughts on what this might mean in practice.

Background

As readers of this blog will know, Stability AI has developed a generative AI tool known as “Stable Diffusion” which creates synthetic images in response to prompts entered by users. Getty alleged that Stable Diffusion was trained using millions of copyright-protected images scraped from its websites without its permission. It originally asserted that those actions infringed its copyright and database rights. Getty also claimed that the outputs (i.e. the images) produced by Stable Diffusion infringed their rights by reproducing substantial parts of their copyright protected works or by bearing Getty’s trade marks (in the form of its watermark). They therefore sued Stability AI for copyright infringement, as well as database right infringement, trade mark infringement and passing off.

At the trial, however, as part of their closing statements, Getty dropped their claims for direct copyright and database right infringement in the context of both the training of Stable Diffusion and the outputs generated by Stable Diffusion. In practice, that meant that the main focus of the case before the court, and yesterday’s judgment, was limited to questions of trade mark infringement and secondary infringement of UK copyright.

Secondary copyright infringement

The majority of the judgment was taken up by the trade mark infringement questions, but it’s fair to say that it was the court’s views on secondary copyright infringement that were the most keenly awaited.

In essence, the question before the court was whether importing (or downloading) a pre-trained generative AI model, like Stable Diffusion, into the UK can amount to secondary infringement of UK copyright. Sections 22 and 23 of the Copyright, Designs and Patents Act 1988 (CDPA) provide that importing or dealing with an article that is, and is known to be, an infringing copy of a work is an act of secondary infringement. Historically, these provisions have only been applied in the context of physical goods and so the first question the court had to consider was whether the word “article” was broad enough to capture intangibles. The court concluded that it was. 

However, the court rejected Getty’s arguments that Stable Diffusion was an “infringing copy”. Whilst section 27(3) of the CDPA says that “[An] article is also an infringing copy if…(b) its making in the United Kingdom would have constituted an infringement of the copyright in the work in question…”, the court found that, by its nature, an infringing copy must be exactly that - a copy. This is where Getty’s claim fell down as the court concluded that Stable Diffusion had never contained or stored copies of any of Getty’s copyright works. It was not therefore an infringing copy. The fact that copies of Getty’s works may have been used in developing that model (in the US) did not change this.

Trade mark infringement

The trade mark infringement side of the dispute has received less press attention, being quite specific to the facts of this case, but Getty were partially successful in their claims on that front (albeit in a limited and historic way) on the grounds of double identity and likelihood of confusion. Getty’s claims of trade mark infringement on the grounds of dilution, tarnishment and unfair advantage, however, all failed, and its claims of passing off were ultimately not addressed.

Comment

When this case first began, IP practitioners across the country were hopeful that it would shed light on whether generative AI developers’ current practices of using unlicensed third party material to train their models was permissible in the UK. Alas, as the trial progressed, it became clear that we wouldn’t get many of the answers we were seeking, with the core questions before the court ultimately being severely limited. In the end, the court wasn’t required to answer any questions about whether use of Getty Images’ copyright works to train Stable Diffusion infringed Getty’s UK copyright (those activities having taken place outside the UK), nor did the court have to address questions about database right infringement or potentially copyright-infringing outputs. 

The fact that Getty’s claim for secondary copyright infringement failed will be a relief to generative AI developers (most of whom are based outside the UK). But it will be seen as a severe blow for holders of UK copyrights who may regard this as opening the door for developers to circumvent their rights by training their models in jurisdictions that are more AI-friendly, before making the trained models available in the UK. Indeed, subject to any changes to the law that may result from the government’s consultation on copyright and AI, the effect of this judgment may be that we see AI companies deliberately avoiding training and developing their models in the UK (which currently looks to be less AI-friendly than other jurisdictions given the limited scope of existing exceptions to primary copyright infringement). In turn, that puts even greater pressure on the outcome of the UK government’s consultation.

We do need to remember, however, that this is only a first instance decision – it remains to be seen if Getty will appeal. It’s also important to bear in mind that this is only one side of the story in this dispute. Getty is running a similar claim in the US, which is where the training and development appears to have taken place. It will be interesting to see how that plays out.