United Kingdom - Getty Images (US) Inc & Ors v Stability AI Ltd

Claimants: Getty Images (US) Inc & Ors (‘Getty’)

Defendant: Stability AI Ltd (‘Stability AI’)

Stability AI made an application to the High Court, seeking reverse summary judgment in its favour and/or strike out of various parts of Getty’s claim. This application was rejected by the High Court and therefore the claims made by Getty (set out in the ‘Summary of key legal arguments’ section below) will be heard at trial. 

By an order of 22 April, a trial on liability (outlined below) is set to be heard between 9 June 2025 to 31 July 2025.

Getty brought proceedings against Stability AI in January 2023. The legal arguments surround the use/reproduction of images from Getty’s websites to train Stability AI’s generative AI tool, “Stable Diffusion”. Stable Diffusion is a deep-learning, text-to-image (and now image-to-image) AI model that is used to generate synthetic images in response to commands entered by users. 

Getty claims that the first iteration of Stable Diffusion was trained using 12m photos, videos and illustrations from Getty’s websites (of which 7.3m are copyright works). It also claims the second iteration was trained using 7.5m of Getty’s photos, videos and illustrations (of which 4.4m are copyright works). 

Stability AI has made its tool available to users in the UK on its website, Dream Studio, where it sells access to compute time for generating images with Stable Diffusion. Stable Diffusion is also available to third parties in the UK to download on an open-source basis.

Fieldfisher is representing Getty and Bird & Bird is representing Stability AI.

Getty’s Claims

1. Copyright. 
Getty is arguing that Stability AI has infringed its copyright works, in particular that:

1.1. in training Stable Diffusion, Stability AI has reproduced the whole or a substantial part of Getty’s copyright works (including by downloading, storing and creating 1000s of altered noisy versions required for the Stable Diffusion model);

1.2. in making Stable Diffusion available to the public in the UK via Dream Studio and on an open-source basis, Stability AI has communicated a substantial part of Getty’s copyright works to the public;

1.3. in making Stable Diffusion available and allowing users to generate AI outputs, Stability AI has authorised reproduction and/or communication to the public by users of Stable Diffusion (where the Stable Diffusion model produces images closely resembling Getty’s copyright works following a user’s prompt); and

1.4. Stability AI possesses in the course of business and/or has offered or exposed for sale or hire an article (Stability Diffusion) which it knows or has reason to believe is an infringing copy of Getty’s copyright works [i.e. an allegation that the model itself is infringing, separate from the works generated by the model]. 

2. Database rights.

2.1. Getty asserts that the substantial investment in its collection of photos, videos and illustrations and associated captions provides it with a database right under the Copyright and Rights in Database Regulations 1997 (and Database Directive).

2.2. Getty is arguing that Stability AI has infringed Getty’s database rights by (1) extracting a substantial part of the contents of the database and/or (2) re-utilising a substantial part of the contents of the database, either in one go or by means of repeated and systematic extraction and/or re-utilisation. 

3. Trade marks. 

Getty is arguing that Stability AI has infringed its trade marks under s.10(1)-(3) Trade Marks Act 1994 (identity, confusion and reputation) by using these signs in the course of trade without consent by virtue of various Getty and ISTOCK trade marks appearing in Stable Diffusion’s output images as a watermark. 

4. Passing off. 

Getty is also arguing that Stability AI’s use of the signs GETTY IMAGES and ISTOCK constitutes a misrepresentation that its synthetic image output is a photo or video owned/licensed by Getty and that Stability AI is therefore passing off itself and/or its goods and services as those of (or as connected/associated to) Getty.

Stability AI’s Defence

1. Copyright. 

Stability AI defends the copyright claims by arguing that:

1.1 any copying for the input (data sourcing and training) stage took place outside the UK;

1.2 any copyright infringement at the output stage results from the user’s actions, not Stability AI’s;

1.3  for communication to the public, there is no appropriate “public”;

1.4  such “infinitesimally small” reproductions of parts of copyright works, as are present in output images of Stable Diffusion, would not amount to a communication of a substantial part of any particular copyright work;

1.5  Stable Diffusion is not an infringing copy (or an article); and

1.6  in any case, Stability AI can rely on a pastiche defence and E-commerce Reg defences (caching and hosting) in respect of AI-generated outputs. 

2.Database. 

Stability AI contests Getty’s database right claim by arguing that:

2.1 a database right does not subsist as the relevant Getty entities do not qualify for protection;

2.2 extraction took place outside the UK;

2.3 there is no re-utilisation of contents of the database as the outputs do not correspond to any specific material from Getty’s database;

3. Trade marks. 

Stability AI argues that:

3.1 AI-generated outputs featuring Getty’s images have only been generated by Getty’s legal team through “wilful contrivance” and are not produced as part of Stability AI’s own commercial communication; and

3.2 there is no likelihood of confusion or unfair advantage/dilution etc. in any case.

4. Passing off. 

Stability AI argues that any misrepresentation is made by the user, not Stability AI, and in any case the AI-generated outputs are not similar to Getty’s trade marks so as to deceive the public.