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

These are proceedings for copyright infringement, database right infringement, trade mark infringement and passing off against the Defendant, an open-source generative artificial intelligence (“AI”) company incorporated in England and Wales in November 2019 with registered offices in London. The claim concerns the Defendant’s deep learning AI model (known as “Stable Diffusion”), which the Defendant makes available to users in the United Kingdom by a variety of means, including via a commercial platform operating under the name of Dream Studio (“Dream Studio”) and by way of download on an open-source basis. Stable Diffusion generates synthetic image outputs in response to commands entered by users.

The First to Fifth Claimants are members of a group of companies (“the Getty Images Group”) ultimately owned and controlled by Getty Images Holdings, Inc. The Getty Images Group is described in the Particulars of Claim as a “pre-eminent global visual content creator and market place”.

The Getty Images Group licenses its content (in the form of millions of visual assets (“the Visual Assets”) including photographs, video footage and illustrations, together with associated captions and key words (“the Captions”) covering a broad range of subject matter) through websites at gettyimages.com, gettyimages.co.uk and istockphoto.com (“the Getty Images Websites”) to creative, corporate and media customers in more than 200 countries. The Visual Assets, which each bear a Getty Images watermark, thus appear in newspapers, magazines, advertising campaigns, films, television programs, books and websites all around the world.

A substantial proportion of the Visual Assets are said to comprise original artistic works and/or film works in which copyright subsists (“the Copyright Works”). The Particulars of Claim assert the existence of millions of Copyright Works.

It is the Claimants’ case that the Visual Assets, the Captions and other information (such as information relating to the creator and/or owner of the rights in relation to the Visual Assets) (“the Data”) are held in a sophisticated database that has been developed by the Getty Images Group since 1995 (“the Database”) and in respect of which a very significant financial investment has been made.

The Visual Assets that appear on the Getty Images Websites are said to be highly desirable for use in connection with AI and machine learning because of their high quality and because they are accompanied by content specific, detailed captions and rich metadata.

The Particulars of Claim were served by the Claimants on 12 May 2023. For the purposes of this hearing, it is common ground that, as alleged in the Particulars of Claim (i) copyright subsists in the Copyright Works and is owned by, or licensed to, the First Claimant on an exclusive basis¹; (ii) the First and/or Second Claimants are the owners of the database right which subsists in the Database both prior to, and after, 31 December 2020 (i.e. the end of the Brexit implementation period); (iii) the First Claimant is the registered proprietor of a number of UK registered trade marks for GETTY IMAGES and the Fifth Claimant is the registered proprietor of a number of UK registered trade marks for ISTOCK in respect of photographs and a variety of digital imaging goods and services (“the Trade Marks”); and (iv) the Claimants have rights in passing off by virtue of the use of the Trade Marks.

In broad terms, the Claimants’ complaint in the proceedings is that the Defendant has “scraped” millions of images from the Getty Images Websites, without the Claimants’ consent, and used those images unlawfully as input to train and develop Stable Diffusion; further that the output of Stable Diffusion in the form of synthetic images (accessed by users in the United Kingdom) is also itself infringing in that it reproduces a substantial part of the Copyright Works and/or bears the Trade Marks.

The Defendant has not yet filed a defence to the claim but has instead made the application to which I shall now turn.

Each party pursues applications made by way of notices dated 28 July 2023 and 25 September 2023 respectively.

The first in time is the Defendant’s application which seeks:

During the course of the hearing, the RFI Application was resolved, with the Claimants agreeing to provide examples of Copyright Works that are alleged to have been infringed by the Defendant. The provision of these examples appears to me to be important in circumstances where no particulars of infringement have previously been given. The Defendant is entitled to know the case it must meet at trial and it is entitled to have an opportunity to respond to that case in its defence. It was agreed between the parties that the examples that would be identified by the Claimants need not be representative and that the production of the examples would not be characterised as “experiments”, such that the details of the internal work carried out in order to produce the examples will be privileged and will not be disclosable in the proceedings. In respect of each of the examples identified, it was also agreed that the Claimants would respond to questions 8-11 and 19-21 of the RFI².

Upon clarification of the Claimants’ case by way of a new proposed amendment to paragraph 50 of the Particulars of Claim, together with a draft Amended Annex 8³, provided on the second day of the hearing, and upon the Claimants confirming that they did not rely for the purposes of advancing their positive case on Louboutin v Amazon (C-148/21 and C-184/21)⁴ the Defendant indicated that it no longer intended to pursue the SJ Application in relation to the Trade Mark and Passing Off Claims. Following the hearing, the Claimants provided a further draft amendment, removing a typo in paragraph 50 as to the year in which the text prompts were inputted and ensuring the addition of wording in paragraph 58 which makes it clear that the case now pleaded in paragraph 50 also applies in respect of the passing off claim. I confirm that permission is granted (as reflected in the Order of 1 November 2023 (“the Order”)) in relation to these amendments. The Defendant’s agreement not to pursue this part of the SJ Application is recorded in the agreed recitals to the Order.

In written submissions on the form of order to be made following the hearing, the Defendant submitted that the SJ Application in relation to the Trade Mark and Passing Off Claims should be adjourned pending the provision of additional information which the Defendant wishes to see in connection with the responses to the RFI. I reject this suggestion. Although the potential to adjourn this aspect of the SJ Application pending the provision of information was floated by Mr Saunders at the hearing, I did not understand it to be pursued following further argument and clarification as to the nature of the information that the Claimants intended to provide. There was no application in relation to the provision of such information. The court has already spent two days hearing the matters at issue between the parties and, if the Defendant wished to make it a condition of withdrawing part of the SJ Application that additional information was required from the Claimants, it should have made that abundantly clear at the hearing. Any such intention is in any event inconsistent with the agreed recitals to the Order, as provided to me in draft by the parties. Once information has been provided by the Claimants in response to the RFI, no doubt there will be scope for the Defendant to make any additional requests that it may consider appropriate. However, I am not prepared to make any order for the provision of additional information at this stage and nor am I prepared to adjourn this aspect of the SJ Application. Accordingly, I make no order on the SJ Application in so far as it relates to the Trade Mark and Passing Off Claims, although the costs associated with that issue will need to be determined at the consequentials hearing.

The SJ Application in relation to the Training and Development Claims and the Secondary Infringement Claims now forms the main battleground on the Defendant’s application.

The Claimants’ application seeks:

It was common ground that, owing to a shortage of time at the hearing, the Fixing Application (which is opposed by the Defendant) should be adjourned to the consequentials hearing following the hand down of this judgment.

The Amendment Application was originally unrelated to, and unaffected by, the SJ Application. However, by reason of the production of a further proposed Amended Particulars of Claim on the second day of the hearing, the scope of the Amendment Application was effectively expanded to include both the amendments to which I have referred in paragraph 13 above and proposed amendments to paragraph 43, the latter being designed to set out in more detail the Claimants’ case on the Training and Development Claim. The Defendant maintains its application for summary judgment on the Training and Development Claim and submits that, even taking these new amendments into account, the Training and Development Claim has no real prospect of success at trial. Accordingly, the Defendant invites me to refuse permission for the proposed amendments to paragraph 43 of the Particulars of Claim. I shall return to address these proposed amendments in the context of dealing with the Training and Development Claim.

The Defendant opposes the Amendment Application in relation to the Original Proposed Amendments for reasons to which I shall return

It is important at this stage to make clear that (in addition to the Trade Mark and Passing Off claims which will now proceed to trial), the Claimants also allege (i) copyright infringement by reason of the copying and communication to the public where Stable Diffusion is used to generate an image that reproduces the whole or a substantial part of the Copyright Works contrary to sections 16(2) and 17 of the Copyright Designs and Patents Act 1988 (“CDPA”); and (ii) database right infringement by re-utilising a substantial part of the Database. Both of these claims rely upon access to Stable Diffusion by users in the United Kingdom. They are not the subject of any applications before the court and (absent further order) will proceed to trial.

There was a substantial amount of evidence served on both sides for this hearing. In so far as this evidence dealt with facts relevant to the SJ Application, it concentrated on the question of whether any of the training and development carried out on Stable Diffusion took place in the United Kingdom. This might ordinarily be regarded as a question that can only be determined at trial, but, it is the submission of the Defendant, that on a proper forensic analysis of the pleaded case and the available evidence, there is nothing whatever to indicate any real dispute of fact.

The Defendant relies upon two statements from Mr Toby Bond, a partner at Bird & Bird, three statements from Mr Peter O’Donoghue, Chief Financial Officer of the Defendant since July 2022, one statement from Mr Mohammed Mostaque, founder and Chief Executive Officer of the Defendant who is based in London, and one statement from Mr Tyler Saltsman, who lives in the US and has worked for the Defendant since May 2023 as Director of HPC Cloud Partnerships (often referred to as the Defendant’s “Head of Supercompute”). Prior to his role at the Defendant, Mr Saltsman worked at Amazon Web Services Inc (“AWS”) for around 3 years.

The Claimants rely upon two statements from Mr Nicholas Rose, a partner at Fieldfisher, together with two expert reports from Professor Hany Farid, a professor at the University of California, Berkeley. Permission was given to both sides to rely upon an expert’s report in the field of the development, training and testing of generative AI models for the purposes of the SJ Application by the court, by consent, on 28 September 2023 (“the September Order”). The Defendant has chosen not to serve, or rely upon, any expert evidence. It is common ground that Professor Farid is a leading expert in AI, albeit that the relevance of his evidence is in dispute. Professor Farid’s first report records that he has been involved in academic research which focuses on digital forensics, forensic science, misinformation, image analysis and human perception. Between 2011 and 2018 he was also involved in a company that was developing software to detect manipulated images.

Pursuant to an application notice issued on the day before the hearing, the Defendant invites me to permit it to rely upon a second statement from Mr Mostaque dated 26 October 2023 and a statement from Mr Rombach, a research scientist at the Defendant between September 2022 and July 2023 and, since then, the Defendant’s Research Director, dated 30 October 2023. Mr Saunders KC, on behalf of the Defendant, submits that these new statements are required to respond to the Claimants’ reply evidence served on 24 October 2023 and to a new point raised in the Claimants’ skeleton argument. He contends that the Claimants are not prejudiced by the introduction of the new evidence, which simply establishes beyond doubt that the Defendant has a clear answer to various factual matters raised by the Claimants.

The Claimants oppose the late introduction of these new statements. In summary, they argue that they are too late, that their timing fails to comply (i) with the provisions of the September Order (to which the parties consented) and also (ii) with the provisions of CPR 24.5(3) which require evidence in reply from an applicant in a summary judgment application to be served on 3 clear days’ notice before the hearing. Ms Lane KC, on behalf of the Claimants, contends that an application for relief from sanctions should have been made by the Defendant and she points out that the Claimants are prejudiced by the late introduction of evidence which they would have wished to consider and investigate in advance of the hearing.

I indicated to the parties that I would read these two statements de bene esse for the purposes of the hearing. Having considered the matter further, in my judgment it would be in the interests of justice to permit Mr Mostaque’s second statement (and in so far as is necessary I grant relief from sanctions in relation to that statement) but Mr Rombach’s statement will be excluded. My reasons are as follows.

The September Order set out a detailed timetable for the service of further evidence relating to the SJ Application. The Defendant had, by that stage, already served its application notice together with the first statements of Messrs Bond and O’Donoghue. The September Order permitted service by the Claimants of the factual evidence on which they wished to rely in response by 25 September 2023, service of any further evidence from the Defendant by 16 October 2023 and service by the Claimants of any evidence in reply to the Defendant’s further evidence by 23 October 2023. There was no provision for a further round of evidence from the Defendant, although paragraph 4 of the order gave liberty to apply to the court for permission to vary the deadlines. The 16 and 23 October deadlines were subsequently shifted by one day by consent between the parties.

The Defendant’s application notice of 30 October 2023, seeking permission to rely upon the new statements, seeks an order “pursuant to paragraph 4” of the September Order, which Mr Saunders characterised as an application to vary that order. However, to my mind, this is not an application to vary the deadlines for the evidence envisaged in that order, but rather an application for a yet further round of reply evidence served less than 3 clear days in advance of the hearing and, in the case of Mr Rombach’s statement, after exchange of skeleton arguments.

Mr Mostaque’s second statement is extremely short and its purpose is to reply to two specific points (made in the second statement of Mr Rose dated 24 October 2023) about six individuals that currently work for the Defendant. The statement was served on 26 October 2023 at 12.13 pm, i.e., just over 36 hours after the Defendant saw the second statement from Mr Rose. The application notice explains that it was prepared as soon as possible, given Mr Mostaque’s existing commitments. Ms Lane did not suggest that there were any further investigations that the Claimants would have wished to make in response to this statement and nor did she suggest that she could not deal with it in her submissions. Accordingly, it seems to me to be in the interests of justice and consistent with the overriding objective that I permit the Defendant to rely upon it. I doubt that relief from sanctions is strictly necessary where the evidence is purely intended as a reply to Mr Rose’s second statement, but, if necessary, I do not regard this as a serious breach of the order and I consider that the need for the statement and its timing has been adequately explained. In all the circumstances it is just to permit reliance upon Mr Mostaque’s second statement.

Although Mr Rombach’s statement is short, it appears to me to fall into a different category. It is said to be designed to address a criticism raised by the Claimants in their skeleton argument as to the lack of evidence produced by the Defendant from individuals in the team which worked on the training and development of Stable Diffusion. However, this criticism was quite plainly made by Mr Rose in his first statement at paragraph 40 and again in his second statement at paragraph 31. If the Defendant wished to address the point, I can see no reason why it did not do so earlier and no reason has been identified. I accept Ms Lane’s submission that if the Claimants had been given more notice of the Defendant’s intention to rely upon evidence from Mr Rombach they would have wished to carry out investigations in advance of the hearing and that the very late service of his evidence has precluded any opportunity of their doing so. In the circumstances, I do not consider it to be consistent with the overriding objective to permit reliance upon his evidence (whether I am being asked to do so by way of variation of the September Order or on the basis of a deemed application for relief from sanctions). I observe in any event that I do not consider that my refusal to permit this evidence in any way affects the outcome of the SJ Application, for reasons which will become clear in due course.

Before turning to the substance of the applications, I should record that a confidentiality order was made on 10 October 2023 (“the Confidentiality Order”) for the purposes of protecting confidential information disclosed in these proceedings on or before the date on which judgment in the Defendant’s application of 28 July 2023 is handed down. This has led to parts of the materials and evidence before me being designated “confidential” or “highly confidential”. It has also led to my being invited to make an order on 31 October 2023 pursuant to CPR 31.22(2) that the evidence served in connection with the application and included in the hearing bundles be treated as confidential and that no use may be made of such documents save for the purposes of these proceedings and in accordance with the terms of the Confidentiality Order. I made the order sought.

For the purposes of this judgment I have tried, in so far as possible, to confine myself to making direct reference only to materials that are not designated as “confidential” or “highly confidential” by the parties. If necessary, this is an issue that may be revisited at the consequentials hearing.

The principles to be applied on an application for summary judgment are largely uncontroversial - the court may grant summary judgment “on the whole of the claim or on an issue”, if it considers that “the party has no real prospect of succeeding on the claim, defence or issue” (CPR r.24.3(a)) and “there is no other compelling reason why the case or issue should be disposed of at a trial” (r.24.3(b)). In determining a summary judgment application, evidence is admissible to establish that the pleaded case is fanciful – albeit that the court will be very cautious about rejecting the claimant’s factual case at the summary stage.

Although they inevitably sought to highlight different aspects, the parties both relied upon the well-known judgment of Lewison J in Easyair Ltd v Opal Telecom Ltd [2009] EWHC 339 (Ch) at [15], approved by the Court of Appeal in AC Ward & Sons Ltd v Catlin (Five) Ltd [2009] EWCA Civ 1098:

“i)

The court must consider whether the claimant has a “realistic” as opposed to a “fanciful” prospect of success: Swain v Hillman [2001] 2 All ER 91;

ii)

A “realistic” claim is one that carries some degree of conviction. This means a claim that is more than merely arguable: ED & F Man Liquid Products v Patel [2003] EWCA Civ 472 at [8];

iii)

In reaching its conclusion the court must not conduct a “mini-trial”: Swain v Hillman;

iv)

This does not mean that the court must take at face value and without analysis everything that a claimant says in his statements before the court. In some cases it may be clear that there is no real substance in factual assertions made, particularly if contradicted by contemporaneous documents: ED & F Man Liquid Products v Patel at [10];

v)

However, in reaching its conclusion the court must take into account not only the evidence actually placed before it on the application for summary judgment, but also the evidence that can reasonably be expected to be available at trial: Royal Brompton Hospital NHS Trust v Hammond (No 5) [2001] EWCA Civ 550;

vi)

Although a case may turn out at trial not to be really complicated, it does not follow that it should be decided without the fuller investigation into the facts at trial than is possible or permissible on summary judgment. Thus the court should hesitate about making a final decision without a trial, even where there is no obvious conflict of fact at the time of the application, where reasonable grounds exist for believing that a fuller investigation into the facts of the case would add to or alter the evidence available to a trial judge and so affect the outcome of the case: Doncaster Pharmaceuticals Group Ltd v Bolton Pharmaceutical Co 100 Ltd [2007] FSR 63;

vii)

On the other hand it is not uncommon for an application under Part 24 to give rise to a short point of law or construction and, if the court is satisfied that it has before it all the evidence necessary for the proper determination of the question and that the parties have had an adequate opportunity to address it in argument, it should grasp the nettle and decide it. The reason is quite simple: if the respondent's case is bad in law, he will in truth have no real prospect of succeeding on his claim or successfully defending the claim against him, as the case may be. Similarly, if the applicant's case is bad in law, the sooner that is determined, the better. If it is possible to show by evidence that although material in the form of documents or oral evidence that would put the documents in another light is not currently before the court, such material is likely to exist and can be expected to be available at trial, it would be wrong to give summary judgment because there would be a real, as opposed to a fanciful, prospect of success. However, it is not enough simply to argue that the case should be allowed to go to trial because something may turn up which would have a bearing on the question of construction: ICI Chemicals & Polymers Ltd v TTE Training Ltd [2007] EWCA Civ 725”.

The overall burden of proof on a summary judgment application rests on the applicant to establish that there are grounds to believe that the respondent has no real prospect of success and that there is no other compelling reason for trial. Once the applicant adduces credible evidence in support of her application, the respondent becomes subject to an evidential burden of proving a real prospect of success. However, the standard of proof by way of rebuttal is not high - the respondent’s case must be more than merely arguable; it must carry some degree of conviction (Easyair at 15(ii) above).

The Defendant also drew my attention to:

Indeed, the need to consider the evidence that might reasonably be expected to be available at trial was recently confirmed in HRH Emere Godwin Bebe Okpabi v Royal Dutch Shell plc [2021] UKSC 3, a case in which the Supreme Court had cause to consider the summary judgment test of real prospect of success. Lord Hamblen JSC pointed out (at [127]) that: “As Lord Briggs JSC stated in Vedanta [202] AC 1045, para 45: ‘the court cannot ignore reasonable grounds which may be disclosed at the summary judgment stage for believing that a fuller investigation of the facts may add to or alter the evidence relevant to the issue’”. Lord Hamblen formulated the question to be posed at [128] as follows: “…are there reasonable grounds for believing that disclosure may materially add to or alter the evidence relevant to whether the claim has a real prospect of success”

On the issue of “compelling reason”, it may be inappropriate to grant summary judgment where similar issues would remain to be determined at a full trial and extensive factual and expert evidence would have to be called, meaning that there would be much less in terms of saving costs and court time than is normal (see Iliffe v Feltham Construction Ltd [2015] EWCA Civ 715 at [71]-[73] per Jackson LJ). However, as the Defendant submitted, the mere existence of other arguable claims which must go to trial cannot, of itself, be a compelling reason why an unarguable claim must proceed to trial.

Once again, the principles are largely common ground between the parties. The court may strike out a statement of case pursuant to CPR 3.4(2)(a), read together with 3.4(1) which makes clear that a statement of case may include part of a statement of case, on the grounds that it “discloses no reasonable grounds for bringing the…claim”. In considering a strike out application, the court must assume that the facts pleaded in the relevant statement of case are true and ask itself whether the claim advanced on the basis of those facts has a real prospect of success.

Practice Direction 3A provides some guidance as to when it may be appropriate to strike out particulars of claim pursuant to CPR 3.4(2)(a), including claims “which are incoherent and make no sense” and claims “which contain a coherent set of facts but those facts, even if true, do not disclose any legally recognisable claim against the defendant”.

Further guidance is found in Begum v Maran (UK) Ltd [2021] EWCA Civ 326, per Coulson LJ at [22]-[24]:

“22.

As to the applicable test itself:

(a)

The court must consider whether the claimant has a “realistic” as opposed to a “fanciful” prospect of success: Swain v Hillman [2001] 1 All ER 91. A realistic claim is one that carries some degree of conviction: ED & F Man Liquid Products v Patel [2003] EWCA Civ 472. But that should not be carried too far: in essence, the court is determining whether or not the claim is “bound to fail”: Altimo Holdings v Kyrgyz Mobil Tel Ltd [2012] 1WLR 1804 at [80] and [82].

(b)

The court must not conduct a mini-trial: Three Rivers District Council v Governor

and Company of the Bank of England (No 3) [2003] 2 AC 1, in particular paragraph 95. Although the court should not automatically accept what the claimant says at face value, it will ordinarily do so unless its factual assertions are demonstrably unsupportable: ED & F Man Liquid Products v Patel; Okpabi and others v Royal Dutch Shell Plc and another [2021] UKSC 3, at paragraph 110. The court should also allow for the possibility that further facts may emerge on discovery or at trial: Royal Brompton Hospital NHS Trust v Hammond (No 5) [2001] EWCA Civ 550; Sutradhar v Natural Environmental Research Council [2006] 4 All ER 490 at [6]; and Okpabi at paragraphs 127-128.

23.

The other principle relevant to the present appeal is that it is not generally appropriate to strike out a claim on assumed facts in an area of developing jurisprudence. Decisions as to novel points of law should be based on actual findings of fact: see Farah v British Airways (The Times 26 January 2000, CA). In that case, the Court of Appeal referred back to the decision of the House of Lords in Barrett v Enfield DC [2001] 2 AC 550 where Lord Browne-Wilkinson said at 557e-g:

“In my speech in the Bedfordshire case [1995] 2 AC 633, 740 – 741 with which the other members of House agreed, I pointed out that unless it was possible to give a certain answer to the question whether the plaintiff’s claim would succeed, the case was inappropriate for striking out. I further said that in an area of the law which was uncertain and developing (such as the circumstances in which a person can be held liable in negligence for the exercise of a statutory duty or power) it is not normally appropriate to strike out. In my judgment it is of great importance that such developments should be on the basis of actual facts found at trial not on hypothetical facts assumed (possibly wrongly) to be true for the purposes of the strike out”.

...

24.

The same point arose more recently in Vedanta Resources PLC & Another v Lungowe & Others [2019] UKSC 20. That was a case where the underlying duty of care was alleged against a parent company, rather than the company involved in the day–to–day running of the mine said to have caused the pollution. Lord Briggs said:

“48.

It might be thought that an assertion that the claim against Vedanta raised a novel and controversial issue in the common law of negligence made it inherently unsuitable for summary determination. It is well settled that difficult issues of law of that kind are best resolved once all the facts have been ascertained at a trial, rather than upon the necessarily abbreviated and hypothetical basis of pleadings or assumed facts.”

Where the strike out plea is based on the nature of the pleading and there is also a summary judgment application made in the alternative which depends purely upon the way in which the case is pleaded, “there is no difference between the tests to be applied by the court under the two rules” (Begum at [20]).

It is common ground that copyright (like database right) is a territorial right which confers protection on its holder only within the territory of the United Kingdom. As Lord Neuberger PSC said in Starbucks (HK) Ltd v British Sky Broadcasting Group plc [2015] UKSC 31 at [59], approving Professor Wadlow in The Law of Passing-off – Unfair Competition by Misrepresentation 4^th ed, 2011, para 3-131: “…the nature of copyright as a legal right of property arising in any given jurisdiction from national legislation, common law or self-executing treaty, means that it must be wrong to speak as if there were a single international copyright”.

The question at the heart of the SJ Application in relation to the Training and Development Claim is whether there is any prima facie evidence that the training and development of Stable Diffusion took place in the United Kingdom (“the Location Issue”). If there is no such evidence, then, submits the Defendant, that claim is bound to fail.

To examine this question, it is necessary to begin by looking closely at the way in which this claim is pleaded in the existing Particulars of Claim.

Paragraph 34 of the Particulars of Claim pleads that “[f]rom a date currently unknown to the Claimants” but prior to the launch of Stable Diffusion version 1.0 on 22 August 2022, “Stable Diffusion 1.0 (and the various sub-versions that were subsequently released) was trained using various subsets of the LAION-5B dataset, specifically LAION-2B-en, LAION-high-resolution, LAION-improved-aesthetics, and LAION-aesthetics v.2 5+.” Paragraph 34 proceeds to define the LAION-5B dataset as the “LAION-5B Dataset” and the subsets, also used to train Stable Diffusion 1.0 as the “LAION-Subsets”.

Paragraph 35 pleads that the LAION-5B Dataset and the LAION Subsets were created by LAION e.V. (“LAION”), an association registered in Hamburg, Germany. It goes on to assert that LAION created the LAION-5B Dataset “with support from the Defendant”.

Paragraphs 36 and 37 plead details as to the LAION-5B Dataset, in particular that it comprises some 5.85 billion CLIP⁵ filtered image text pairs, created by “scraping links to photographs and videos, together with associated captions, from the web” including from the Getty Images Websites. The pleading describes the links in the LAION-5B Dataset and the LAION-Subsets that match to Content on the Getty Images Websites as “the Scraped Links” and asserts that the Claimants have identified “around 12 million” of these links, obtained without their consent. Paragraph 38 pleads that Stable Diffusion 1.0 was trained using around 12 million Visual Assets (of which around 7.3 million are Copyright Works), together with associated Captions, from the Getty Images Websites.

Paragraph 39 refers to the launch of Stable Diffusion 2.0 on or about 24 November 2022 and paragraph 40 pleads that “[f]rom a date currently unknown to the Claimants” but prior to that date, Stable Diffusion 2.0 was trained “using the LAION-5B Dataset…”. Paragraph 41 pleads that Stable Diffusion 2.0 was trained using around 7.5 million Visual Assets (of which around 4.4 million are Copyright Works) together with associated Captions, from the Getty Images Websites.

Paragraph 47 of the Particulars of Claim sets out in some detail the steps that it is alleged the Defendant took to train Stable Diffusion, albeit there is no allegation as to where they took place. These steps are accepted by the Defendant as accurate for the purposes of the SJ Application and I need not set them out in detail here. Suffice, for present purposes, to observe that the training process is said to involve numerous iterative steps involving the creation and storage of content in various different forms. The Defendant has not sought to address the detail of the training process in its evidence.

Paragraph 43 of the Particulars of Claim pleads as follows:

“The Defendant’s development team includes a significant number of individuals who reside and work in the UK, who will be identified upon receipt of appropriate confidentiality undertakings, including the Defendant’s lead generative AI developer. It is to be inferred that, during the development and training of Stable Diffusion, visual assets and associated captions, used for the development and training of Stable Diffusion, including Visual Assets (including Copyright Works) and Captions from the Getty Images Websites, were downloaded on servers and/or computers in the UK”.

Pausing there, I make the following observations about this pleading:

Paragraph 52.1 of the Particulars of Claim pleads that the Defendant has infringed the copyright in the Copyright Works by:

“As regards the acts identified above at paragraphs 43 and 47, reproducing the whole or a substantial part of the Copyright Works or each or any of them, contrary to section 17 CDPA…”.

Paragraph 55.1 pleads that, further or alternatively, the Defendant has infringed the database right in the Database by:

“As regards the acts identified above at paragraphs 43 and 47, extracting a substantial part of the contents of the Database contrary to Article 7(1) of the Directive and Regulation 16(1) of the Regulations…”.

In its skeleton argument, the Defendant emphasised that the First Claimant has already brought a parallel claim against it (and its US parent company Stability Inc), commenced in February 2023, in the United States District Court for the District of Delaware. The Delaware proceedings make substantially the same allegations concerning the training and use of Stable Diffusion, albeit they allege that the Defendant “maintains cloud computing and physical server resources in the United States” and that by training Stable Diffusion, the Defendant has engaged in “infringing conduct in the United States and in Delaware”. The Defendant said in its skeleton that it is “striking” that the Claimants’ resistance to the SJ Application “is premised on the contrary assertion that training of Stable Diffusion took place in the UK”. However, given the way in which paragraph 43 of the Particulars of Claim is pleaded, there need be no inconsistency between the allegations in the Delaware proceedings and the allegations in this case and Mr Saunders accepted during his oral submissions that the existence of the Delaware proceedings is in no way determinative of the application before this court; it is not inconsistent to allege that training occurred in the United States but that during that process (or indeed during an earlier stage of development) Copyright Works were downloaded on servers and/or computers in the UK.

Against that background, and in summary only, the Defendant contends that its evidence establishes that: (i) all of the computing resources used by the Defendant for its training of Stable Diffusion have, at all times, been located outside the United Kingdom; in particular, all of the training “compute” infrastructure is located in two USA datacentres operated by AWS; (ii) although services have been provided to the Defendant in the United Kingdom by AWS, they have not related to the training or development of Stable Diffusion; and (iii) none of the Defendant’s employees based in the United Kingdom has ever worked on (or had the right equipment to work on) developing or training Stable Diffusion. Mr Saunders described the Defendant’s factual case as being that it had “built one of the world’s largest supercomputers based in the AWS cloud” and that no work had been done as part of that process in the United Kingdom. Accordingly, the Defendant submits that it is safe for the court to conclude that all acts of training and development took place outside the United Kingdom and that, accordingly, the Training and Development Claim is bound to fail.

Furthermore, the Defendant contends that, on a proper analysis of the evidence, there is no factual dispute identified by the Claimants that matters for the purposes of the Location Issue; on the contrary, much of the Claimants’ case is little more than, in Mr Saunders’ words, “kicking up dust”. Mr Saunders goes so far as to say that the court could accept all of the Claimants’ evidence but still conclude that; (i) it does not contradict the Defendant’s evidence; (ii) much of it is purely speculative and therefore (iii) it is irrelevant to the Location Issue.

It is not possible for me in this judgment to go through all of the detailed factual evidence set out in each of the many statements on which the Defendant relies. However, focussing specifically on the critical points in the evidence to which my attention was expressly drawn by the Defendant, I can summarise as follows:

Pausing here, I observe that if this were the trial of this action, the evidence to which I have referred above would (on its face) provide strong support for a finding that, on the balance of probabilities, no development or training of Stable Diffusion has taken place in the United Kingdom. This is not, however, the trial of this action and if I am to grant summary judgment I must be satisfied that there is no real prospect of the Claimants being able, at trial, to refute this evidence and to establish on balance that there are grounds for the inference that they invite the court to draw at paragraph 43 of the Particulars of Claim (as now amplified by proposed amendments to which I shall return). Further, I must be satisfied that there are no reasonable grounds for believing that a fuller investigation into the facts would add to or alter the evidence available to the trial judge and so affect the outcome of the Location Issue.

Having examined with care all the evidence before the court, I am not so satisfied. There seems to me to be (i) evidence potentially pointing away from the factual determination on the Location Issue that I am invited to reach by the Defendant; (ii) evidence raising unanswered questions and inconsistencies relevant to that determination; and (iii) reasonable grounds for believing that disclosure may add to or alter the evidence relevant to the question of where the training and development of Stable Diffusion took place. All of this clearly supports the proposition that the Training and Development Claim has a real prospect of success and must be permitted to go to trial. The Location Issue is certainly not an issue on which I can say at present that the Claimants’ claim is doomed to fail.

In this context, it is worth observing that the Claimants do not hold, or have access to, any of the relevant information on the central question of whether any development and training may have taken place in the United Kingdom and so could not sensibly be expected at this stage (at least until disclosure, when the position may change) to plead anything other than a purely inferential case. Whilst the inference pleaded at paragraph 43 of the Particulars of Claim appears somewhat weak (reliant as it is in the original pleading purely upon the location of the Defendant’s development team), it has now been supplemented in the proposed amended pleading by additional factual material which, in my judgment, when taken together with the evidence and the arguments to which I shall now turn, is sufficient to raise a realistic, as opposed to a fanciful, case. I would have held that this was the position on the pleading as it stands (given the evidence), but the amended pleading has, to my mind, substantially strengthened that position.

It is not for the court to conduct a mini-trial on an application of this sort and so it does not appear to me to be appropriate to conduct a detailed analysis of the evidence to which I was taken at the hearing and the arguments on that evidence. However, having regard to the central points only, my reasons for concluding that the Training and Development Claim has a real prospect of success at trial are as follows:

It is common ground that the Defendant is based in the UK, has its principal place of business in the UK and has (and had both prior to and after March 2022) a number of UK-based employees in its development team. Notwithstanding Mr Mostaque’s evidence that none of these employees worked on the training and development of Stable Diffusion, there are contemporaneous documents which appear to tell a different story. Furthermore, to my mind, these documents raise the spectre that Mr Mostaque’s evidence is either inaccurate, or incomplete; at the very least they suggest a conflict of evidence and thus reasonable grounds for believing that a fuller investigation into the facts of the case would add to or alter the evidence available to the trial judge in respect of the Location Issue. Thus:

In addition, I accept the Claimants’ submissions that given the central importance of the question of whether any of the UK based team worked on Stable Diffusion, Mr Mostaque’s own evidence in his statements remains incomplete and continues to give rise to queries which cannot be determined on an application of this sort. In so far as he has given evidence about his investigations into UK employees on his development team (identified as a subset of a list of 38 employees and bearing job titles which could suggest that they worked on Stable Diffusion), he has not contacted any of those employees directly (but has done so only via his HR team or his lawyers). Having explained this procedure, he has not confirmed that responses were even received from each employee, let alone identified exactly how they responded. In other words, he has not explained the source of his stated “confidence” that none of these employees has worked on developing or training Stable Diffusion. Mr Saunders’ submission that his evidence is satisfactory because it sets out “the confirmations given” appears to me, on analysis, to be incorrect. I reject the Defendant’s submission that Mr Mostaque’s evidence as to these UK based employees is “clear” or definitive.

Had I permitted Mr Rombach’s statement into evidence I would have observed that it is wholly insufficient to put the genie back into the box. It refers only to employees; it makes no mention of independent contractors, of “the LAION guys”, the Russians or the Ukrainians. It does not descend into the details of the names, dates or locations of individual employees and would have carried little weight in all the circumstances. Whilst it would have remedied the absence of evidence from anyone admittedly involved in the development and training of Stable Diffusion, it would not have cleared up the issues to which I have already referred; it would certainly not have left me in a position where I could properly determine that the Claimants have no prospect of success on the Training and Development Claim at trial. Importantly, Mr Rombach’s statement did not even attempt to explain exactly what the development and training of Stable Diffusion actually involved, evidence which, to my mind, is crucial background to a proper determination of the Location Issue.

In this context I also note Professor Farid’s unchallenged evidence to the effect that the development and training of a generative AI model is a complex, iterative process requiring input from a team of individuals over a relatively long period of time. Whilst this tells me little about what the Defendant actually did, I accept that (absent challenge from the Defendant) it supports the real possibility (when seen in conjunction with the other evidence to which I have referred) that human intervention of one kind or another may have occurred through the Defendant’s employees or independent contractors operating in the United Kingdom, whether at the stage of development or training.

I should add that I am not deflected from this view by the content of the LMU press release on which Mr Saunders relied in his responsive submissions, which records, as at 1 September 2022, that The Machine Vision & Learning Group led by computer scientist Bjorn Ommer “has developed one of the most powerful image synthesis algorithms in existence”, a reference to the “new Stable Diffusion AI model”. Although this press release goes on to say that “[t]he underlying algorithm was developed” by the Machine Vision & Learning Group, it also says “[i]n their project, the LMU scientists had the support of the start-up Stability.Ai, on whose servers the AI model was trained”. The question of what this support involved and whether it went beyond providing access to its AWS cloud resources, lies at the heart of this application. It is not, however, a question that I can determine on a summary basis.

The Claimants’ evidence as to computer resources again appears, on close analysis, to raise legitimate (and relevant) factual questions which cannot be resolved without further investigation.

Professor Farid opines in his first report that the development of Stable Diffusion “may or may not have used AWS services and may have included using and evaluating data in the UK as well as moving data into the UK either on desktops or via other cloud services depending on where [the Defendant’s] development team and their management was located”. Professor Farid (quite properly) accepts that he cannot say where, or how, Stable Diffusion was developed but he observes that “it would have been possible to train the open-source version of Stable Diffusion on a desktop computer, and so it is feasible, and indeed likely (given the huge costs involved in using the powerful cloud services) that a combination of local laptop/desktop devices and AWS cloud services was used to develop the model…”. Mr Saltsman rejects the proposition that the development of Stable Diffusion would have been more cost-effective if done on local machines and so rejects Professor Farid’s evidence, but he does not say in terms that local machines were not used in the development of Stable Diffusion and he was not, in any event, employed by the Defendant prior to May 2023. Mr Mostaque’s evidence is that the Defendant’s employees who accessed the AWS cluster to work on Stable Diffusion will have done so from local machines although he says those individuals were all outside the UK. However, given the holes in his evidence as to individuals who might have worked on Stable Diffusion, there appears to me to be a prospect, which is more than merely fanciful, that local machines were used during this process in the UK.

The list of computer assets produced by Mr O’Donoghue goes no further back than assets purchased in February 2022; it says nothing about any assets that may have been purchased by the Defendant prior to that date. Mr O’Donoghue says in his third statement that the Defendant’s assets do not include server racks containing GPUs and that the Defendant’s UK employees are not provided with standard laptops or high-powered workstations containing “expensive GPUs”. Professor Farid confirms in his second report that he cannot say what assets the Defendant may have had, but he reiterates the view expressed in his first report that “much of the early part of the development process would not have required the computing power offered by cloud services” (i.e. the AWS cluster) and that many of the standard laptops that the Defendant’s UK employees had “would have been suitable for the early part of the training and development of Stable Diffusion”, further that they “will have contained GPUs”. This evidence does not appear to be directly challenged and it is not addressed by Mr Saltsman’s evidence to the effect that his work on the AWS cluster involved the use of specialist GPUs. I note also that the Defendant does not challenge Professor Farid’s evidence that, from his own experience “it is common practice” for employees of technology start-ups to use their personal computers for work purposes. Whilst, of course, Professor Farid cannot say what resources the Defendant had, his evidence (when taken together with the evidence to which I have already referred) raises the genuine possibility that at an early stage of development and training, including when the Defendant was offering “support” to the LMU scientists, local UK computing resources may have been used by employees or independent contractors not yet addressed (adequately or at all) in the Defendant’s evidence.

In all the circumstances I cannot determine, as Mr Saunders’ invites me to do, that all of the evidence points solely to development and training taking place on the AWS cloud in the US.

Furthermore, and importantly in my judgment, there is unexplained evidence of data transfers to London from AWS in the United States:

Pausing there, I have now addressed the evidence on which the Claimants seek to rely in putting forward their proposed amendments to paragraph 43. I agree with Ms Lane that these amendments provide a stronger factual premise from which to invite the court to draw the inference pleaded and I cannot determine that they have no real prospect of success. No criticism is made in respect of the way in which these proposed amendments are pleaded. Accordingly, I am prepared to grant permission for these amendments.

The matters to which I have already referred are, in my judgment, sufficient to determine that the Claimants have a real prospect of success on the Training and Development claim at trial. However, Ms Lane also relied upon conflicts of fact on the evidence, which I accept provide an additional reason why this claim is unsuitable for summary judgment; they are not conflicts which can be determined at this hearing.

I have already identified some of these conflicts, but I note in particular:

Finally, Ms Lane relies upon issues in respect of which there are reasonable grounds for believing that a fuller investigation into the facts of the case would add to or alter the evidence available to the trial judge relevant to the Location Issue. Once again, this follows on from many of the points I have already made, but I accept that in summary it is reasonable to assume that evidence will be available at trial to shed light on:

I reject the Defendant’s submission that these enquiries are unnecessary (essentially because the evidence is clear and there is no reasonable prospect of the factual position relating to the location of the training and development of Stable Diffusion changing further to such enquiries). For all the reasons that I have already given, I cannot determine that the Defendant’s evidence is clear such that the Training and Development claim is manifestly unsustainable. I also cannot conclude, as the Defendant invited me to do, that a trial on the Location Issue would be “pointless and wasteful” or that, if the matter needs to go to trial, that should relate only to a limited subset of the Location Issue (as Mr Saunders submitted in his reply).

For all the reasons set out above, I am unable to conclude that the Training and Development Claim carries no degree of conviction and so has no real prospect of success. I dismiss the SJ Application in relation to this claim.

As developed in the Defendant’s skeleton argument, the application in relation to the Secondary Infringement Claim raised a number of arguments. Ultimately, however, as Mr Saunders accepted in his oral submissions, it really stands or falls on one point of law, namely the true interpretation of the word “article” in sections 22, 23 and 27 of the CDPA. In particular, whether sections 22 and 23 CDPA are limited to dealings in “articles” which are tangible things or whether they may also encompass dealings in intangible things (such as making available software on a website).

The Defendant argues that there is no real prospect of the Claimants establishing at trial that an “article” is an intangible thing and so no real prospect of them succeeding in the claims of secondary infringement at paragraphs 52.4, 52.5, 53 and 54 of the Particulars of Claim. These paragraphs refer back to paragraphs 44-46, which in summary plead the launch of the Defendant’s website through which it sells access to Dream Studio, thereby enabling users to generate images with Stable Diffusion, the availability of Dream Studio to users in the United Kingdom, and the fact that the Defendant also makes Stable Diffusion available to download on an open-source basis to third parties in the United Kingdom via GitHub.

Paragraphs 52.4 and 52.5 are in the following terms:

“52.4

As regards the acts identified above at paragraphs 44 to 46, importing into the United Kingdom, otherwise than for private and domestic use, an article, namely Stable Diffusion, which is and which the Defendant knows or has reason to believe is, an infringing copy of the Copyright Works or each or any of them, contrary to section 22 of the CDPA; and/or

52.5

As regards the acts identified above at paragraphs 44 to 46, possessing in the course of a business, selling or letting for hire, or offering or exposing for sale or hire, an article, namely Stable Diffusion, which is and which the Defendant knows or has reason to believe is, an infringing copy of the Copyright Works or each or any of them, contrary to section 23 of the CDPA”.

Paragraph 53 pleads that “pending disclosure and/or evidence” the Claimants’ case that Stable Diffusion is an infringing copy of the Copyright Works is dependent either upon the fact (paragraph 53.1) that the making of Stable Diffusion constituted an infringement (i.e. the Development and Training Claim) or the fact that (paragraph 53.2) Stable Diffusion “has been imported into the United Kingdom” and by reason of the manner in which it was trained as set forth in paragraph 47 of the Particulars of Claim “its making in the United Kingdom would have constituted an infringement” of the copyright in the Copyright Works.

Paragraph 54 pleads particulars of knowledge, said by the Defendant in its skeleton argument to be insufficiently particularised, but not pursued at the hearing.

The key provisions of the CDPA are in the following terms:

“22.

Secondary infringement: importing infringing copy

The copyright in a work is infringed by a person who, without licence of the copyright owner, imports into the United Kingdom, otherwise than for his private and domestic use, an article which is, and which he knows or has reason to believe is, an infringing copy of the work.

23.

Secondary infringement: possessing or dealing with infringing copy

The copyright in a work is infringed by a person who, without the licence of the copyright owner-

(a)

Possesses in the course of a business,

(b)

Sells or lets for hire or offers or exposes for sale or hire,

(c)

In the course of a business exhibits in public or distributes, or

(d)

Distributes otherwise than in the course of a business to such an extent as to affect prejudicially the owner of the copyright, an article which is, and which he knows or has reason to believe is, an infringing copy of the work.

…

27.

Meaning of “infringing copy”

(1)

In this Part “infringing copy” in relation to a copyright work, shall be construed in accordance with this section.

(2)

An article is an infringing copy if its making constituted an infringement of the copyright in the work in question

(3)

An article is also an infringing copy if-

(a)

it has been or is proposed to be imported into the United Kingdom, and

(b)

its making in the United Kingdom would have constituted an infringement of the copyright in the work in question, or a breach of an exclusive licence agreement relating to that work.

…”

It is common ground that section 27(2) requires that the making of the article “constituted an infringement of the copyright in the work” which, pursuant to sections 1(2) and 178 (definition of copyright) must be taken as a reference to United Kingdom copyright and therefore to acts in the United Kingdom. It follows from this that the making of an article outside the United Kingdom is not within section 27(2). Thus, in so far as the Claimants rely upon section 27(2) in paragraph 53(1) of the Particulars of Claim, their claim of secondary infringement will be dependent upon the outcome of the Training and Development Claim and I am not concerned with that here.

Section 27(3), on the other hand, relates to an article made outside the United Kingdom which has been, or is proposed to be, imported into the United Kingdom. Such an article can also be treated as an infringing copy only if both of the conditions in 27(3) are met: to succeed, it is therefore essential that it either has been, or is proposed to be, imported. For the purposes of the Claimants’ claims under sections 22 and 23, the acts of importation, possession or dealing must be by the Defendant.

There is no relevant statutory definition of “article”. However, the Defendant’s case is that, on a true interpretation of the statutory provisions to which I have referred, an article can only be something which is capable of being “imported” (section 22); “possessed” (section 23(a)); “distributed” (section 23(c)-(d)) and “made” (section 27), i.e. a physical tangible thing. The Defendant’s skeleton notes that the only cases that counsel has been able to identify under sections 22 and 23 all concern physical tangible articles, such as pairs of shoes (LA Gear Inc v Hi-Tec Sports plc [1992] FSR 121); compact discs (Springsteen v Flute International Ltd [1999] EMLR 180) and football sticker albums (FAPL v Panini UK Ltd [2004] 1 WLR 1147).

The Defendant points out that it is settled law that abstract information is not property and cannot be the subject of a claim for detinue or conversion, relying in particular upon the following observations made by Floyd LJ in Your Response Limited v Datateam Business Media Limited [2014] EWCA Civ 281 at [42]:

“An electronic database consists of structured information. Although information may give rise to intellectual property rights, such as database right and copyright, the law has been reluctant to treat information itself as property. When information is created and recorded there are sharp distinctions between the information itself, the physical medium on which the information is recorded and the rights to which the information gives rise. Whilst the physical medium and the rights are treated as property, the information itself has never been.”

and by Arnold LJ in Thaler v Comptroller General of Patents, Trade Marks and Designs [2021] EWCA Civ 1374⁶ at [125]:

“There are good reasons of principle and policy for this rule: it would have alarming consequences if there was property in information”.

Thus, submits the Defendant, whilst cases relating to whether there is property in information do not bind this court when it comes to the interpretation of sections 22 and 23 of the CDPA, those cases suggest that it is necessary to look to dealings in tangible property (goods or merchandise) which embody information when assessing the “article” for the purposes of secondary infringement. The Defendant contends that examples of this approach may be seen in Sony Computer Entertainment Inc v Ball [2005] FSR 9 and in Wheat v Google LLC [2018] EWHC 550 (Ch), to which I shall return in a moment.

Finally, the Defendant contends that this question of statutory construction is a straightforward question of law, that the Claimants have no real prospect of establishing that “article” includes intangible articles and that the point should be determined (whether in its favour or not) at this hearing.

Having considered all the arguments with care, I decline to take the course suggested by the Defendant. In essence I consider there to be good reasons to postpone to trial a decision on the true statutory interpretation of the word “article”.

First, I am not satisfied on the arguments that I have heard that the Claimants are bound to fail on this point. I say that for the following reasons:

Second, and following on from the points I have already made, this argument appears to me to raise a novel question, not definitively determined previously, which would be better resolved once all the facts have been ascertained at trial rather than upon the necessarily abbreviated and hypothetical basis of pleadings or assumed facts. I reject the suggestion by the Defendant, for reasons I have given, that the point was effectively determined in Sony v Ball and I also reject the suggestion that it is safe to rely on the decision in Wheat v Google. The question of whether an “article” for the purposes of sections 22 and 23 of the CDPA can include an intangible article, has not to my mind been previously argued in detail or decided by this court. It is not straightforward and it arises in connection with a case of secondary infringement involving internet downloading, which (with the exception of Wheat v Google) has not previously been considered. In my judgment it would be infinitely preferable for the court to make findings of fact as to the precise nature of the acts said to give rise to the secondary infringement (for example precisely how Stable Diffusion has been imported into the UK) before going on to consider the legal question.

Third, I note Ms Lane’s submission that there are good policy reasons why “article” should be read as meaning both tangible and intangible articles. She points out that on the Defendant’s construction, there would be an infringing copy and secondary infringement if the Defendant had brought a copy of Stable Diffusion into the UK on a memory stick but, because of developments in technology which mean this can now be achieved via a cloud-based service, there would be no infringement. Whilst I make no decision on the impact of this point (which was not accepted by Mr Saunders who maintains, essentially, that if Parliament had intended to include intangible information within the meaning of “article” it would have made that clear), it is yet another factor which appears to me to support the decision I have reached.

For all the reasons given, I dismiss the SJ Application in relation to the Secondary Infringement Claim and decline to decide the point of law identified by the Defendant. That point of law must be decided by the trial judge upon full and comprehensive argument from both sides and further to factual findings as to the nature of the acts said to give rise to the secondary infringement.

The Claimants’ Amendment Application seeks to introduce a new case in respect of an image-to-image feature which enables Stable Diffusion to generate a synthetic image output in response to an image uploaded by a user (either with or without a text prompt). The user is able to determine how closely the synthetic image output matches the image prompt by the use of an “image strength” slider, with the maximum image strength and/or values approaching the maximum image strength providing images which comprise the whole, or a substantial part, of the image prompt.

The amendment sought introduces claims of copyright infringement, database right infringement, trade mark infringement and passing off, in relation to the image-to-image feature. The relevant paragraphs are 50A-50B.3 of the draft Amended Particulars of Claim:

“50A. From a date unknown but since at least in or around March 2023 the Defendant

introduced an image-to-image feature which enables Stable Diffusion to generate a synthetic image output in response to an image uploaded by a user (either via Dream Studio or via a copy of Stable Diffusion downloaded from GitHub or via commercial offerings using Stable Diffusion), either with or without a text command. The user is able to determine how closely the synthetic image output matches the image prompt by the use of an “image strength” slider, with the maximum image strength and/or values approaching the maximum image strength providing images which comprise the whole or a substantial part of the image prompt.

50B. Accordingly, when a user provides an image prompt (either via Dream Studio or via a copy of Stable Diffusion downloaded from GitHub or via commercial offerings using Stable Diffusion) comprising one of the Copyright Works or the Visual Assets and uses the maximum image strength or a value approaching the maximum image strength (the precise value to be identified by way of disclosure and/or evidence), the synthetic image output comprises the whole or a substantial part of the Copyright Work or Visual Asset. Pending the provision of disclosure and/or evidence, the Claimants rely upon the following facts and matters:

50B.1. In the case of Stable Diffusion downloaded from GitHub, the image strength is a value between 0.0 and 1.0 that controls the amount of noise that is added to the image prompt. As explained in the guidance on the GitHub website (a copy of which is attached hereto at Annex 8A) values that approach 1.0 allow for lots of variations in the synthetic image output by removing all pixel-based information. By contrast, values that approach 0.0 (i.e. the maximum image strength) result in a synthetic image output that reproduces the whole or a substantial part of the image prompt.

50B.2. When an image is uploaded by a user, it is first converted into noise, and this noise pattern is used as an input to the image synthesis process. Thus, the image-to-image feature uses both Stable Diffusion and the pixels of the user-supplied, uploaded image.

50B.3. Copies of synthetic images that were generated in response to image prompts

comprising a Copyright Work or Visual Asset, together with the image strength indicated, uploaded via a copy of Stable Diffusion downloaded from GitHub are attached hereto at Annex 8B”.

Cross-references to these paragraphs then appear in amended versions of paragraphs 52.2, 52.3 and 55.2 making express the plea of copyright infringement (by communication to the public or by authorising acts of reproduction or communication to the public, contrary to sections 16(2) and 20 CDPA) and database right infringement (by re-utilising a substantial part of the contents of the Database). Additional amendments to paragraphs 33 and 39 plead that the use of Stable Diffusion to generate synthetic images in response to image prompts commenced on “a date unknown but since at least in or around March 2023”.

The Defendant opposes these proposed amendments on two grounds: first that they have no real prospect of success and second that they are inadequately particularised.

The court has a broad discretionary power to grant permission to amend pursuant to CPR 17.3. The principles applicable to amendments are not in dispute. They were summarised by Carr J in Quah Su-Ling v Goldman Sachs International [2015] EWHC 759 (Comm) at [36]-[39], albeit primarily in the context of late amendments, and by Lambert J in Pearce v East and North Hertfordshire NHS Trust [2020] EWHC 1504 (QB) at [10]. In short, and drawing upon the test for summary judgment, if amendments are to be permitted they must have a realistic prospect of success and carry a degree of conviction; it is not enough that the amendment is merely arguable (see Kawasaki Kisen Kaisha Ltd v James Kemball Limited [2021] EWCA Civ 33 per Popplewell LJ at [18(1)]). The court is entitled to refuse an amendment which raises a version of events “which is inherently implausible, self-contradictory or is not supported by contemporaneous documentation” (see Collier v P&MJ Wright (Holdings) Ltd [2007] EWCA Civ 1329). The pleading must be coherent and properly particularised (see Kawasaki at [18(2)]).

The overriding objective is of “central importance”; thus amendment applications will always involve the court striking a balance between injustice to the applicant if the amendment is refused and injustice to the opposing party and other litigants in general if the amendment is permitted.

This was the final application to be heard by the court and was argued more sparingly than the other two applications. However, it became clear during the course of argument that the Defendant’s objections to the Original Proposed Amendments were something of a moveable feast. The Defendant’s skeleton argument focussed on reasons why they were “fundamentally unclear” (many of which have since fallen away) and devoted only two paragraphs in a 48 page skeleton to the argument that the Original Proposed Amendments disclose no real prospect of success – effectively on the grounds that the Claimants have accepted that they generated the synthetic images in Annex 8B themselves such that it cannot be said that there was any act of infringement without the Claimants’ consent. However, I did not understand this point to be maintained at the hearing. Instead, the Defendant’s oral submissions took a rather different form, as I set out in more detail below. This was unfortunate, because it meant that new points were developed at some speed by Mr Saunders on his feet and had to be responded to in the same manner (and without notice) by Ms Lane.

Doing the best I can to make sense of the arguments as they developed, the Defendant’s key argument appears now to be that when a copy of Stable Diffusion is downloaded by a user from GitHub with a view to then entering the image prompt, it is in fact the user who brings about the copying (i.e. by voluntarily entering into Stable Diffusion a Copyright Work that the user has downloaded from the Getty Images Websites) and not the Defendant. If the image strength slider is fixed to produce maximum image strength, or values approaching the maximum image strength (as the Original Proposed Amendments assume at 50A) then the output image is obviously very close to the image that has been entered – this, says the Defendant, “is no more than an allegation that you end up with what you started with and has little to do with the Defendant’s software” because the generational process has been dialled down to the lowest possible setting. Furthermore, the download of Stable Diffusion to GitHub takes place on the user’s own computer and the software then responds to instructions from the user. This, says the Defendant, cannot possibly amount to infringement on the part of the Defendant; it is taking place locally and cannot therefore amount to a communication to the public.

If, on the other hand, Stable Diffusion is accessed via Dream Studio, then while the Defendant accepts that there is no local download (because the image will be uploaded onto the Dream Studio platform), the Defendant says that it is very difficult to see how it is doing anything more than providing a service by making the Dream Studio platform available – i.e. it is acting merely as an intermediary or host in the sense that there is no active behaviour on its part. Furthermore, the Defendant says that the claim involving Dream Studio is arbitrable under the Dream Studio terms of service and is not within the court’s jurisdiction.

The Defendant makes similar points in relation to the database right infringement claim, asking rhetorically in oral submissions how an image entered by a user can involve an extraction of the Database by the Defendant.

The Claimants respond that the Original Proposed Amendment is clear that although the original image will have been obtained by the user and is then entered into the Stable Diffusion software (just as the text prompt is entered by the user) the synthetic output in the form of an image is the result of differing levels of corruption by noise; it is this output that has been authorised, and communicated to the public, by the Defendant, regardless of whether the user has accessed Stable Diffusion via Dream Studio, by downloading a copy from GitHub or via commercial offerings using Stable Diffusion. The Claimants point to Professor Farid’s evidence as to the way in which the synthetic image is created, which is unchallenged. As for the additional point made in relation to Dream Studio, the Claimants say that although they have referred to Dream Studio in the Original Proposed Amendments, and although the pleading is obviously intended to apply to any version of Stable Diffusion, howsoever offered, the particulars that they have chosen to provide relate to GitHub, thereby avoiding the jurisdictional issue which it is accepted arises in relation to Dream Studio.

Ms Lane explained during her submissions that, although not directly relevant to the question of whether the image-to-image case has a real prospect of success, the vice in respect of which the Claimants are seeking protection is more than merely hypothetical. Ms Lane points out that Stable Diffusion can provide a means by which a user can obtain an image that is almost exactly the same as a Copyright Work but without being required to pay for it, thereby facilitating unfair competition with the Claimants’ Copyright Works which have involved acts of intellectual creation.

Having regard to the arguments as they were presented in court, I consider that the image-to-image claim as pleaded (and subject to the additional points I raise below) has a real prospect of success and that, having regard to the balancing exercise to which I have referred, it is plainly a claim that should go to trial. My reasons are as follows.

The claim of copyright infringement as advanced is based on more than the Defendant merely acting as intermediary, or host (cf. Google France SARL v Louis Vuitton Malletier SA C-236/08 EU:C:2010:159 at [55]-[58] in the context of trade mark infringement). As is pleaded in 50B.2, and as is confirmed by Professor Farid, an image which is entered into the Stable Diffusion software is ingested into the model, is pixelated, made into a noisy image and is then used to create part of the output. In other words, the software is reacting to, and engaging with, the image that is entered by the user, thereby undertaking a process which leads to the generation by the software of output in the form of an image. This factual case, which is said to apply regardless of whether the image strength is set to a value that allows for lots of variations in the synthetic image output or whether it is set to a value approaching the maximum image strength, is not challenged by the Defendant.

Upon close analysis of the technology at trial, the factual position may prove to be different. It may be possible to establish that, as Mr Saunders submits, the user who inputs the image (particularly where the image strength slider is set to produce a maximum image strength) is “doing the interaction” and that Stable Diffusion is really playing an immaterial part. However, I cannot possibly determine that factual technical issue at present and I note that in its skeleton argument the Defendant was unable to go further than to say that its software “has little to do” with the creation of the output image, a tacit admission that it at least has something to do with it. Resolution of this issue will require the court to hear detailed evidence as to the workings of the Stable Diffusion software, including the extent to which (if at all) the positioning of the image strength slider affects the way in which the software functions. For the purposes of the amendment application, I certainly cannot determine at this stage that the claim as pleaded has no real prospect of success.

Furthermore, the points I have made at 109 and 110 above appear to me to place this claim (at least on the facts as presently pleaded) in very similar territory to the text-to-image claim. Where there is no challenge to the text-to-image claim and it is accepted by the Defendant that that claim must go to trial, I agree with Ms Lane that it would be rather odd if the court were to find itself in a position where it could not also determine the image-to-image claim, which relies upon the same technology and a very similar mechanism for generating images – the only difference being the nature of the prompt required for the generation of the image. I disagree with Mr Saunders that the proposed amendments necessarily concern a “fundamentally different kind of case”, or, at least, that that is a conclusion that the court is able to reach at this stage.

As for the question of whether the “local” nature of the activity (when there is a downloading of Stable Diffusion from GitHub onto a user’s personal computer) negates any prospect of the Claimants establishing a case of infringement, the Claimants say that they have a sufficiently arguable case of (i) communication to the public (which they say includes communication to a “new public” in the form of users; i.e. not the public that the Claimants originally intended to have access to its Copyright Works) together with (ii) authorising reproduction by users. In this regard:

Given that the Claimants say that every version of Stable Diffusion that is distributed to UK users is infringing, whether it is via Dream Studio or GitHub, I do not consider there to be any reason for present purposes to require particulars of the claim in so far as it concerns Dream Studio prior to granting permission. Particulars in relation to downloads via GitHub should suffice. Accordingly, the jurisdictional issue raised by the Defendant takes matters no further.

The claim of database right infringement, as I understand it, raises the same issues; certainly the parties did not address me by way of any different arguments. In focussing upon the “extraction” of the Database, the Defendant appears to have overlooked the fact that this claim is pleaded (in this context) at paragraph 55.2 of the original Particulars of Claim as involving the re-utilisation of a substantial part of the contents of the Database. This plea was originally the subject of the RFI which prompted clarification from the Claimants to the effect that the re-utilisation involved the provision of access to Stable Diffusion. Ms Lane explained this during her submissions on the RFI Application as re-utilisation by doing the acts pleaded in paragraphs 44-46 and 48 to 49 of the PoC, in other words these were the acts by which the Database was made available to the public. I did not understand Mr Saunders to pursue his request for further particulars in relation to this, or indeed to maintain that this plea should be struck out. There is no reason to suppose, given the analysis I have set out above, that a different position pertains to the database right infringement claim now that the cross reference to paragraphs 50A and 50B has been included in paragraph 55.2 of the Original Proposed Amendments and accordingly I reject the Defendant’s case that this claim has no real prospect of success for all the reasons I have identified in connection with the copyright infringement claim.

For the sake of completeness, I record that:

Prior to circulation of a draft of this Judgment to the parties, I have determined the final issues arising in relation to the Order and my reasoning in relation to those matters is set out in the early part of this judgment.

For reasons set out above, I intend to dismiss the SJ Application and permit the amendments to the Particulars of Claim identified in this judgment subject to sight of a final draft Amended Particulars of Claim dealing with the additional points I have raised.

Accordingly, I invite the Claimants to produce a final form of Amended Particulars of Claim for review in advance of the consequentials hearing. In light of my judgment, I do not expect the terms of the Amended Particulars of Claim to be controversial.

The Sixth Claimant is included as a representative party of the parties that have concurrent rights of action with the First Claimant under ss 101 and 102 CDPA 1988, pursuant to CPR 19.6. The Defendant reserves its position as to the way in which this representative claim is put, but it was not a live issue at the hearing.

After the hearing there was a dispute between the parties over the terms of the order reflecting this agreement. The Claimants were concerned that the Defendant was seeking in the draft order to preserve its position to argue later that it would be entitled to disclosure relating to the exercise undertaken by the Claimants to produce the example synthetic outputs for the purposes of responding to the RFIs. This would not have reflected the concession the Defendant made at the hearing and I do not consider that it is now in a position to try to resurrect any such entitlement. However, my understanding of the Defendant’s position in light of further short written submissions is that it takes the view that any decisions concerning experiments and disclosure going beyond the process required for the purposes of responding to the RFI Application will be matters for the CMC at a later stage. This does not appear to me to cut across its concession made at the hearing. The final Order dated 1 November 2023 but sealed on 23 November 2023 reflects these points.

A second dispute arose after the hearing as to the scope of an Amended Annex 8 to be provided by the Claimants which will include examples of the synthetic output relied upon using a prompt. At the hearing, the Claimants indicated that they would provide examples in relation to text prompts which “are not exactly the Getty Images caption”. However, the Defendant then stated that if the Claimants are seeking to maintain a case that the use of different words nevertheless prompts infringing output “it would be helpful to have an example of that”. I understood this to be accepted by Ms Lane, albeit that the transcript does not record her physical acknowledgment. Accordingly, the Order reflects the requirement to provide examples using a prompt which is not the whole of a Getty Images Caption, but also adds the requirement (in so far as the Claimants seek to maintain a case that words other than words used in the Getty Images Caption prompt the production of infringing output) to provide an example of the same.

A third issue arose between the parties on the draft Order following the hearing as to the scope of the Claimants’ agreement in this regard. I understood it to be the case that the Claimants were confirming at the hearing that they did not advance a positive case arising by reason of the decision in Louboutin v Amazon. I did not understand them to be accepting that they could never again refer to that case in the context of, for example, responding to any pleading from the Defendant. I amended the draft Order to reflect this position. Given the extent to which the parties have disagreed on numerous issues in these proceedings, I do not consider that this is an issue that should be left to be addressed, if necessary, by reference merely to the transcript of the hearing.

Contrastive Language-Image Pre-training

I understand this case to have been heard on appeal in the Supreme Court, albeit not on this point.

Warner Music UK Ltd v Tunein Inc [2021] RPC 11 per Arnold LJ at [70]. The summary of CJEU case law in this paragraph runs to 15 sub-paragraphs. What appears from the summary is that ‘communication to the public’ must be interpreted broadly and, further, that determining whether there has been a communication to the public requires an individualised assessment based on various different factors, including the factual circumstances in which the works have been made available to recipients together with the number of persons who have had access to the works. Assuming these to be relevant factors under domestic law, they will be matters for evidence at trial; I cannot see how I can be satisfied by reference purely to this authority that there is no real prospect of the Claimants succeeding on the image-to-image claim. Further and in any event, I note that paragraph [70] is not part of the ratio of this decision – neither Rose LJ, nor Sir Geoffrey Vos MR agreed with it at paragraphs [181] and [191] respectively and neither of them considered it helpful to provide such a summary where it was unlikely to be valid for long and where the restatement of CJEU principles in domestic judgments were not strictly on point. To my mind this merely serves to emphasise that this court is in no position to determine that the Claimants’ case of communication to the public has no prospect of success.