Getty Images (US) Inc & Ors v Stability AI Ltd [2025] EWHC 38 (Ch)

[2025] EWHC 38 (Ch)Case No IL-2023-000007IN THE HIGH COURT OF JUSTICEBUSINESS AND PROPERTY COURTS OF ENGLAND AND WALESINTELLECTDUAL PROPERTY (ChD)Venue Royal Courts of Justice,, Rolls Building, Fetter Lane,London,EC4A 1NLDate 14/01/2025
Before
MRS JUSTICE JOANNA SMITH DBE
Between
STABILITY AI LTDDefendant
Between
GETTY IMAGES (US) INCGETTY IMAGES INTERNATIONAL UCGETTY IMAGES (UK) LIMITEDGETTY IMAGES DEVCO UK LIMITEDISTOCKPHOTO LPTHOMAS M BARWICK INC
Ms Lindsay Lane KC, Ms Jessie Bowhill and Mr Joshua Marshall (instructed by Fieldfisher LLP) for ClaimantsMr Nicholas Saunders KC, Mr Henry Edwards and Mr Edward Cronan (instructed by Bird & Bird LLP) for DefendantHearing Hearing dates: 14, 15 and 26 November 2024
Approved JudgmentThis judgment was handed down remotely at 10.00 am on 14 January, 2025 by circulation to the parties or their representatives by e-mail and by release to the National Archives.
[1]This is my reserved judgment in respect of an application made by the Defendant at a CMC held on 14, 15 and 26 November 2024. The application, issued on 13 June 2024 (“the Application”), is for an Order that pursuant to CPR r.19.8(2) the Sixth Claimant may not act as representative for the class of individuals identified in paragraph 17 of the Amended Particulars of Claim dated 19 January 2024 (“the APoC”).

Background to the Application

[2]To put this Application in context, I need to set out something about the proceedings themselves and the issues to which they give rise. The next few paragraphs are slightly adapted from my judgment in an unsuccessful summary judgment/strike out application made by the Defendant in 2023 (Getty Images (US) Inc v Stability AI Limited [2023] EWHC 3090 (Ch)).[3]The proceedings are 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. Stable Diffusion generates synthetic image outputs in response to commands entered by users.[4]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 Amended Particulars of Claim as a “pre-eminent global visual content creator and market place”.[5]The Getty Images Group licenses its content (in the form of millions of visual assets including photographs, video footage and illustrations, together with associated captions, covering a broad range of subject matter) (“the Content”) through websites to creative, corporate and media customers in more than 200 countries (“the Getty Images Websites”).[6]A substantial proportion of the Content is said to comprise original artistic works and/or film works in which copyright subsists (“the Copyright Works”). The APoC asserts the existence of millions of Copyright Works in respect of which it is alleged that the First Claimant is either the copyright owner or the exclusive licensee.[7]In broad terms, the Claimants’ complaint 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, the Claimants say that the output of Stable Diffusion is itself infringing, not least because it is said to reproduce a substantial part of the Claimants’ Copyright Works and/or bears the Claimants’ trade marks.[8]The Defendant admits that “at least some images from the Getty Images Website were used during the training of Stable Diffusion” but it has not identified those images.[9]The Claimants raise various causes of action, but for present purposes I need concentrate only on their claims of copyright infringement (“the Copyright Infringement Claims”), which are advanced on three bases:a. first, a claim of copyright infringement arising from the Claimants’ allegation that during the development and training of Stable Diffusion, the Content including the Claimants’ Copyright Works, was downloaded on servers and/or computers in the United Kingdom (“the Training and Development Claim”);b. second, a claim of secondary infringement of copyright said by the Claimants to arise by reason of the importation of an “article”, namely the pre-trained Stable Diffusion software, into the UK (“the Secondary Infringement Claim”); andc. third, a claim 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 (“the Output Claim” The Output Claim also involves a claim of database right infringement in relation to the Database (as defined in the APoC) by way of certain synthesised image outputs of Stable Diffusion. However, for present purposes the reference to the Output Claim is only to the Copyright Infringement Claim as identified here. ).[10]The Output Claim is advanced in respect of(i) the generation of images from text prompts (“thetext prompts claim”);(ii) the generation of images from image prompts (“theimage prompts claim”); and(iii) the generation of images from combined image and text prompts (“the image plus text prompts claim”), the latter being a very recent addition to the Claimants’ pleaded case.[11]In view of the vast number of Copyright Works alleged to be involved in the Copyright Infringement Claims, the Claimants rely upon a selection (“the Sample Works”) for the purpose of establishing subsistence and ownership. These comprise Works A to K. Works A to D are works that are owned by the First Claimant. Works E to K are works that are owned by the Sixth Claimant and exclusively licensed to the First Claimant. In its Amended Defence, the Defendant confirms that it is “content that the question whether its activities have involved infringement of the Copyright Works be resolved by reference to the Sample Works” (whilst also noting that it does not accept any entitlement to general or wider relief in the event the court were to make a finding of infringement). Further, “[i]n the interests of procedural economy”, the Defendant admits the allegations of authorship and subsistence of UK copyright in relation to the Sample Works.[12]In respect of the Copyright Infringement Claims the Claimants seek by way of relief, inter alia:(i) a declaration that copyright subsists in the Copyright Works and has been infringed by the Defendant;(ii) an injunction to restrain the Defendant (without a written licence or agreement from the Claimants) from acting in various ways which would infringe copyright in the Copyright Works; and(iii) an inquiry as to damages for copyright infringement or, at the election of the Claimants, an account of profits.[13]It is the Claimants’ case that there are likely to be in excess of 50,000 photographers and Content contributors who are owners of the copyright subsisting in Content that has been licensed on an exclusive basis to the First Claimant over several decades. By reason of these exclusive licences, each of these copyright owners is alleged to have a concurrent right of action with the First Claimant under sections 101 and 102 of the Copyright, Designs and Patents Act 1988 (“the CDPA”) to restrain the Defendant’s unlawful acts of copyright infringement and to seek relief in respect of such acts.[14]Against that background, the action is also brought by the Sixth Claimant, Thomas M. Barwick, Inc, (a company incorporated under the laws of Washington and engaged in the process of creating and licensing visual content) in a representative capacity, as explained in paragraph 17 of the APoC The representative claim has been included in the Particulars of Claim since 12 May 2023. . A prominent photographer, Mr Thomas Barwick, is the Sixth Claimant’s director. Mr Barwick has had a long professional relationship with the Getty Images Group and many thousands of his images and videos are available on the Getty Images Websites. It is the Claimants’ case that Mr Barwick has assigned all legal and equitable title in the works created by him to the Sixth Claimant which has in turn exclusively licensed Content to the First Claimant. Attached to the APoC at Annex 2 are details of various agreements entered into between the Sixth Claimant and the First Claimant. In its Amended Defence, the Defendant admits the pleaded licensing arrangements relating to the Sixth Claimant (save that it denies that the pleaded chain of title is effective in passing title to any Copyright Works from the Second to the First Claimants). It makes “no wider admissions as to other Copyright Works, and whether the Sample Works are representative of other Copyright Works which may be within the First Claimant’s portfolio of images and films”.[15]The Claimants plead in paragraph 18 of the APoC that the Sixth Claimant “represents and has the same interest in this claim as the parties who are owners of artistic works and film works that have been licensed on an exclusive basis to the First Claimant” (“the Represented Parties”).[16]The Claimants further particularised their representative claim in a Response to a Request For information dated 30 May 2023 (“the Further Information”) in the following terms: “…the class of persons represented by the Sixth Claimant are those who are owners of the copyright subsisting in artistic works and film works that have been licensed on an exclusive basis to the First Claimant, the copyright in which has been infringed by the Defendant. The persons that fall within such class can be identified on the basis that(i) they have entered into an exclusive licence with the First Claimant in respect of artistic works and/or film works; and(ii) the exclusively licensed works include works which were used to train Stable Diffusion, as alleged in paragraph 34 of the Particulars of Claim”.[17]The Claimants assert that it would be disproportionate to identify all members of the class owing to their enormous number and/or the fact that “the precise works used to train Stable Diffusion are within the knowledge of the Defendant”. They go on to state in the Further Information that the interest shared by the Sixth Claimant and the Represented Parties is that of “ensuring that their copyright is not infringed by the Defendant”. The Represented Parties are said to be entitled to: “…all those remedies relating to copyright claimed in the prayer for relief of the Particulars of Claim, including an inquiry as to damages, on the basis of the Defendant’s acts of copyright infringement in respect of works in which the Represented Parties own the copyright and have licensed the same on an exclusive basis to the First Claimant”.[18]Finally, the Further Information states that the question of whether any individual assessment of damages will be necessary is “a matter that will fall for determination following the trial on liability and/or at the damages inquiry or account”.[19]In its Amended Defence, the Defendant pleads that the Claimants’ original pleading involved an unclear and imprecise definition of the represented class. In broad terms, the Defendant denies that the Sixth Claimant shares “the same interest” as the other alleged owners of copyright in Content exclusively licensed to the First Claimant and asserts that identifying whether their Content has been used to train Stable Diffusion, whether their copyright has been infringed and whether they are entitled to damages “requires an individualised assessment”. The Defendant denies that the action can proceed on a representative basis.[20]In an Amended Reply, the Claimants deny that they have failed to set out proper particulars of the basis for the representative claim, given the content of the Further Information. They admit that the Represented Parties “do not all have licensing terms which are identical to each other in every respect”, but they assert that “many of the Represented Parties have entered into standard licence terms (known as ‘the Contributor Agreement’)” and that although the Contributor Agreement has varied over time “in all versions of the Contributor Agreement the relevant content is licensed to the First Claimant exclusively on a worldwide basis to be marketed and sublicensed for any purpose”. Further, the Claimants assert that “each of the Represented Parties have the same licensing terms in the respect which is material for the purpose of CPR 19.8, that is that the relevant content is licensed to the First Claimant exclusively in (at least) the UK”. If there is any diversity of interest between the Represented Parties (which the Claimants deny) they assert that it may be resolved, if necessary “as part of any inquiry as to damages or account of profits”.[21]In addition to the fact that standard licence terms upon which the Claimants rely provide for exclusivity, in their evidence on the Application, the Claimants also point out that those standard licence terms also give to the First Claimant the right to bring proceedings for infringement of the licensed Content (“the Right To Control Claims Clause”).

Procedural Developments

[22]As they were entitled to do, the Claimants made the representative claim from the outset of the proceedings. The Defendant put that claim in issue as I have already identified but, thereafter, case management proceeded without any further real focus on the claim of the Sixth Claimant or the potential ramifications for the trial of a representative claim. As will become apparent, I consider this to have been a mistake. Where a representative claim is to be pursued it appears to me to be critical that its scope and likely impact on the trial, together with any directions which may be required in order to flush out the issues on that claim, is considered in detail from an early stage of the proceedings.[23]In July 2023, the Defendant applied for summary judgment and/or strike out in respect of various elements of the claim, an application which I heard on 1 November 2023. At that application, the Defendant reserved its position as to the representative claim but did not seek to raise it as a live issue. The Claimants also did not seek to raise it for consideration. The summary judgment application was subsequently dismissed. Since then, the court has heard various applications and made a number of orders as to the management of the case.[24]On 22 April 2024, Master McQuail ordered a first trial to determine liability which has since been listed with a time estimate of 18 days (including four non-sitting days) to commence in a 5 day window starting on 9 June 2025 (“the First Trial”).[25]The Defendant eventually made the Application on 13 June 2024. It was supported by five short paragraphs in the application notice explaining that the Defendant would rely upon its pleading, but that any hearing of the Application should take place only following the provision by the Claimants of a fully particularised case on the Copyright Infringement Claims, at which point the court would then be able to determine “whether the sixth claimant is capable of acting as a representative party”. Accordingly the Defendant sought further directions from the court.[26]The matter came back before me on 9 July 2024 for a CCMC. In an Order of the same date (“the CCMC Order”), I gave directions for the service of evidence in the Application from the Claimants, including copies of “all versions” of the Contributor Agreement referred to in the Amended Reply and copies of any other licensing terms which were common to more than 500 of the Represented Parties.[27]The CCMC Order also addressed various outstanding issues of case management, including granting permission for amendments to be made to the APoC (a Re-Amended Particulars of Claim (“the RAPoC”) was subsequently served on 12 July 2024). The CCMC Order recognised (in the recitals) the need for the parties to correspond with a view to “resolving the position regarding whether the liability trial in these proceedings on the issue of subsistence and ownership of copyright in the Copyright Works” would proceed on the basis of “the Sample Works The Sample Works were unchanged in the RAPoC. ” and any Copyright Works relied upon by the Claimants in a Statement of Case on Infringement (“the SOCI”). The CCMC Order provided for the SOCI to be served on 30 August 2024 (albeit that the SOCI was in fact served on 6 September 2024 and updated on 13 September 2024). The SOCI was required in order to support “the breadth of [the Claimants’] case on the Output Claims”.[28]A further hearing took place before me on 20 September 2024 designed to deal with a specific issue that had arisen as to the disclosure of information relevant to Amazon Web Service data transfers into the UK, together with an application by the Defendant for disclosure of images and prompts used or generated by the Claimants in the preparation of the SOCI. As reflected in the Order of the same date, the application for disclosure of these images and prompts was adjourned, with the Claimants voluntarily agreeing to provide the majority of the disclosure of images and prompts sought. Disclosure was then provided on 25 September 2024.[29]On 26 September 2024, the Claimants served the eighth witness statement of Nicholas Rose (“Rose 8”) containing their reply evidence in the Application. Specifically Mr Rose sets out some background to the Claimants’ licensing business and this claim, some background to the Sixth Claimant and a brief description of the Contributor Agreements and other agreements that were being disclosed by the Claimants in accordance with the CCMC Order. This evidence provides additional detail about licensing agreements to which the Sixth Claimant is party over and above the agreements already pleaded by the Claimants.[30]On 21 October 2024 the Defendant served the sixth witness statement of Toby Bond (“Bond 6”) in response to the evidence in Rose 8. Mr Bond addresses the procedural background to the Application, the issue of class definition, the interests represented by the members of the represented class and the case management implications of the Claimants’ proposed representative claim.[31]On 4 November 2024, the Claimants served further evidence in the form of the ninth witness statement of Mr Rose (“Rose 9”). In this statement, Mr Rose referred to a number of additional licence agreements which had been identified after service of Rose 8 and had not previously been listed in Annex 2 to the RAPoC. These included Barwick Agreement #9, also referred to by the parties as Contributor Agreement #17 (“CA#17”), dated 3 October 2023. The Claimants maintain that CA#17 is the most relevant of the licence agreements owing to the fact that it has superseded all previous agreements. I was taken to its terms during the course of the hearing.[32]On the same day, the parties gave Extended Disclosure in the proceedings in accordance with the CCMC Order, having agreed by consent to extend the deadline.[33]Over the last couple of months, the Claimants’ case, as set out in the SOCI, has been amended with a view to clarifying exactly how the Claimants intend to advance their case on the Output Claims at the First Trial. The up to date position (only really clarified at, or shortly before, the hearing) is that the Claimants now identify(i) 17 Copyright Works on which they rely (alone) for the purposes of the text prompts claim These were first said to be “representative” of a wider infringement, then they were said to be “examples” and now it is accepted that the 17 Copyright Works are the only works on which the Claimants intend to rely to prove the text prompts claim. ;(ii) 5 Copyright Works (a subset of the original 17) on which they rely (alone) for the image plus text prompts claim; and(iii) 5 Copyright Works (again a subset of the original 17) on which they rely as “examples” for the image prompts claim. These “examples” are said to be “representative of the infringing Output that will be obtained” if using the relevant image prompt. The First Claimant is alleged to own outright 11 of these identified Copyright Works, while the remaining 6 Copyright Works are said to have been authored by a number of different individuals who subsequently entered into an exclusive licence agreement with the First Claimant. None of these individual licensors is the Sixth Claimant.[34]Prior to the clarification of the Claimants’ case referred to above, the Defendant served a responsive SOCI on 8 November 2024 which complained as to inconsistencies between the Claimants’ case as set out in correspondence and in its SOCI. The Defendant’s responsive SOCI has inevitably now been somewhat overtaken by events.[35]On 3 December 2024, after the hearing in the Application, the Claimants served a Re-Re-Amended Particulars of Claim (“the RRAPoC”) together with an Amended SOCI reflecting their clarified case. Permission for the RRAPoC was granted by Order dated 6 December 2024. Amongst other things, the RRAPoC now asserts that the Sixth Claimant entered into licence agreements with the First and/or the Fifth Claimant and it identifies the forms of agreement that were in force at the date of the acts complained of. Amendments are made to Annex 2 to identify the additional agreements on which the Claimants now seek to rely. Re-Re-Amended Annex 8G to the RRAPoC relies on a few additional synthetic images generated in response to image prompts, a couple of which appear to have been authored by the Sixth Claimant. The Defendant has permission under the Order of 6 December 2024 to serve a Re-Amended Defence and Re-Amended Reply SOCI by 23 December 2024.[36]The Order of 6 December 2024 expressly provides that following service of the Defendant’s amended pleadings, the parties shall correspond “with a view to resolving the Samples Question in relation to the output claim alleging infringement by the use of image prompt”. The need for this only became apparent at the hearing of the Application in circumstances where the Claimants indicated an intention to amend their case (as is now done at 50B of the RRAPoC) to plead (a narrower) image prompts claim by reference to the use of an image strength of “a value between 0.0 and 0.25”, a pleading to which the Defendant wishes to respond before seeking to address the question of sampling.[37]Pausing there, although there has already been a need for a number of hearings in this matter to address the numerous procedural issues on which the parties have been unable to agree, nevertheless these have taken place in circumstances where directions have already been given for the progress of this case to the First Trial in approximately 6 months’ time. Owing to the court dealing with specific disputes between the parties and often having insufficient time to address all of those disputes, there has to date been no consideration by the court as to the possible impact on the First Trial of(i) the continuing development of each side’s positions on sampling for the purposes of the Copyright Infringement Claim; and(ii) the representative claim. The chronology to which I have referred shows that the Claimants did not particularise their Output Claim until service of the SOCI and that, even then, the way in which they have sought to advance that claim (specifically as to how it is to be proved at trial) has been something of a moveable feast. No doubt these issues are, at least in part, a function of the very considerable substantive and procedural challenges that present themselves in litigation involving allegations of infringement on such a massive scale.[38]The Claimants contend that none of this matters, because the issues that have largely been agreed between the parties for determination at the First Trial are unaffected by any individual issues that might arise in respect of the Represented Parties. They point out that the List of Issues for trial prepared for the CCMC does not include any issues relating to authorship, subsistence or ownership of the Copyright Works and that no disclosure has been sought by the Defendant in relation to these issues. The List of Issues also raises no issue in relation to the assertion that the Sixth Claimant entered into an exclusive licence with the First Claimant and the Amended Defence does not suggest that any individualised assessment is required in relation to either ownership or the existence of an exclusive licence – a clear indication, submit the Claimants, that there is really no dispute on these topics.[39]The Claimants further submit that it would be impossible to join all of the Represented Parties to these proceedings and that if the court were to accede to the Defendant’s Application that would be “tantamount to finding that infringement on a huge scale by using copyright works for an AI model cannot be litigated in the English courts because there is no mechanism to deal with the enormous number of works and parties involved”. The Claimants contend that this is precisely the kind of situation in which a claim brought by a representative party pursuant to CPR 19.8 was intended to be used. Further they say that their intention to rely upon samples for certain of their Copyright Infringement Claims has been clear for some time, that that intention is consistent with the pursuit of a representative claim and they reject any suggestion that there are case management reasons for acceding to the Application.[40]Alternatively, and as a fall-back position, the Claimants contend that they should be given permission pursuant to CPR 19.3(1) for these proceedings to be brought without joinder of the exclusive licensors.[41]The Defendant profoundly disagrees with the position adopted by the Claimants. It submits that the Claimants bear the burden of advancing their representative claim and that this ought to have involved the identification of proposals at an early stage for its determination. The Defendant contends that this has not happened, that there remain various outstanding issues as to how the Copyright Infringement Claims are to be tried (including, most significantly, as to sampling in respect of various of those claims) and that, accordingly, if the representative claim is permitted to continue, the court will be left with “an untriable mess”. The Defendant strongly objects to the proposed fall-back position under CPR 19.3 which has never been pleaded and was raised for the first time only in one paragraph at the end of Rose 9.

The Hearing of the Application

[42]I heard the Application over the course of two days on 14 and 15 November 2024 (“the Hearing”). During the course of the argument it became clear that insufficient focus had been brought to bear on the efficient management of the Copyright Infringement Claims to the First Trial, including the extent to which sampling of Copyright Works under the various Copyright Infringement Claims (including the representative claim) would be appropriate. The List of Issues produced for the CCMC does not identify any issues at all arising in connection with the Sixth Claimant and none has been proposed by the Claimants. Furthermore, as the submissions developed, the extent of the disagreement between the parties as to how the Copyright Infringement Claims were to be dealt with at the First Trial became starkly apparent.[43]This unsatisfactory state of affairs led to proposals for the management of the First Trial being advanced on the second day of the Hearing by the Defendant, some of which appeared capable of agreement, whilst others quite obviously needed further discussion. Also on the second day, the Claimants indicated that it might be appropriate further to amend their RAPoC on the image prompts claim (an amendment which has since been permitted in the RRAPoC as explained above).[44]Given the importance of the court having a clear understanding of each side’s proposals for the management of the claim (including the proposed representative claim) to trial, I suggested that the Application be adjourned, that the parties attempt to agree in so far as possible upon how each of the Copyright Infringement Claims should be dealt with at the First Trial and that the parties then return for a third day of argument.[45]In advance of the resumed hearing on 26 November 2024 (“the Resumed Hearing”), the parties liaised over their individual proposals for the First Trial but were regrettably unable to reach any real agreement; indeed in a number of respects agreements apparently made in court at the Hearing appear to have been abandoned. The parties duly produced further skeleton arguments addressing their individual contentions for the management of the Copyright Infringement Claims at trial, together with their submissions as to the impact of their respective proposals on the Application. Each party made (very different) proposals as to the Orders the court should make for the determination of the Copyright Infringement Claims at the First Trial. These proposals appeared to shift over the course of the Resumed Hearing and I therefore asked the parties to produce further drafts after the hearing clarifying their individual proposals as to the form of Order the court should make. It will be necessary for me to return to these in due course.[46]For now, suffice to say that it appeared from the Defendant’s skeleton argument for the Resumed Hearing that the Defendant was floating the possibility that the court deal only with the Getty-owned copyright claims at the First Trial and ‘park’ the representative claim for determination at a later date. This proposal, unheralded as it was and following two full days of court time spent arguing about the Application, came as something of a surprise. The Claimants had been given no opportunity to consider it in advance of the Resumed Hearing and the Defendant had not even attempted properly to explain its proposal in its skeleton argument, let alone produce a draft Order setting out that proposal. On his feet, Mr Saunders KC initially appeared to confirm that the Defendant considered a pragmatic way forward to involve dealing at trial with only what he described as “big picture points” arising in respect of the Getty-owned Copyright Works. However, following a break during which I permitted him to take instructions on what was to happen to the Application in light of this submission, Mr Saunders confirmed that the Defendant still wanted the court to determine the Application, albeit that the Defendant wished to emphasise its position that the representative claim raises complex and difficult issues which would complicate the management of the trial and which are generally unlikely to add much, if anything, to the court’s determination of the core disputes arising in this case.[47]I make the preliminary observation that, whilst the way in which these submissions developed at such a late stage was unfortunate, I have considerable sympathy with the Defendant on its underlying point. The issues in this case are highly complex and it remains a challenge on both sides to work out and to plead how they are to be addressed at trial. The Defendant was right to say in its skeleton for the Resumed Hearing that the dominant concern for the parties and the court must be to arrive at a selection of triable issues which will resolve as much of the commercial dispute between the parties as is practically possible at the First Trial in a way that is both proportionate to the dispute and fair to the court and other users of the court system. If the Defendant had made this proposal a great deal earlier it may have been possible to manage the case accordingly, including by making provision for a means of dealing with the exclusively licensed Copyright Works after an initial determination of the “big picture points”.[48]Instead, however, the court must now try to deal swiftly with the difficult Application it has before it in circumstances where it often appeared during both the Hearing and the Resumed Hearing that the parties were working out their positions on the Application and on the management of the case to trial on the hoof. With only around 6 months to trial and with the parties’ pleaded cases still subject to change and further clarification, not to mention the host of disagreements as to case management generally, I consider this to be both unfortunate and unsatisfactory.[49]In producing this judgment I have read and had regard to all of the parties’ submissions and evidence and I have, as necessary, revisited the daily transcripts. For obvious reasons, I have not dealt with all of those arguments (many of which fall by the wayside in light of the decisions I have arrived at) but I have considered them and taken them all into account.

Relevant Legal Principles

[50]Section 101 of the CDPA provides as follows:
“Rights and remedies of exclusive licensee. (1) An exclusive licensee has, except against the copyright owner, the same rights and remedies in respect of matters occurring after the grant of the licence as if the licence had been an assignment. (2) His rights and remedies are concurrent with those of the copyright owner; and references in the relevant provisions of this Part to the copyright owner shall be construed accordingly. (3) In an action brought by an exclusive licensee by virtue of this section a defendant may avail himself of any defence which would have been available to him if the action had been brought by the copyright owner.”
[51]An exclusive licence is defined in section 92(1) of the CDPA as:
“A licence in writing signed by or on behalf of the copyright owner authorising the licensee to the exclusion of all other persons, including the person granting the licence, to exercise a right which would otherwise be exercisable exclusively by the copyright owner.”
[52]Thus, the essential requirement under section 92(1) is that the licence operates to the exclusion of all other persons (including the grantor) in respect of a right that would otherwise be exercisable exclusively by the copyright owner.[53]Section 102 of the CDPA provides that:
“Exercise of concurrent rights (1) Where an action for infringement of copyright brought by the copyright owner or an exclusive licensee relates (wholly or partly) to an infringement in respect of which they have concurrent rights of action, the copyright owner or, as the case may be, the exclusive licensee may not, without the leave of the court, proceed with the action unless the other is either joined as a plaintiff or added as a defendant. (2) A copyright owner or exclusive licensee who is added as a defendant in pursuance of subsection (1) is not liable for any costs in the action unless he takes part in the proceedings.”
[54]Section 102 of the CDPA is echoed by CPR r.19.3 which provides that:
“(1) All persons jointly entitled to the remedy claimed by a claimant must be parties unless the court orders otherwise. (2) If any person does not agree to be a claimant, he must be made a defendant, unless the court orders otherwise.”
[55]There are, in effect, three regimes under CPR Part 19 under which multiple claimants may bring claims in one claim form and one set of proceedings: joinder (rr. 19.1 and 7.3), group litigation (r.19.21-24), and, as in this claim, representative proceedings brought under CPR r.19.8. It is of course not impossible to join very large numbers of individual claimants to one consolidated particulars of claim – they can be identified in a schedule and the claim can then be case managed sensibly so that sufficient sample claimants and claims may be determined at trial for extrapolation purposes (see Jalla v Shell International Trading and Shipping Co Limited [2021] EWCA Civ 1389, per Coulson LJ at [67]). However, an alternative approach (subject to satisfying the relevant requirements) is to bring a representative claim under CPR r.19.8.[56]CPR r.19.8 provides (emphasis added): “(1) Where more than one person has the same interest in a claim –(a) the claim may be begun; or(b) the court may order that the claim be continued, by or against one or more of the persons who have the same interest as representatives of any other persons who have that interest. (2) The court may direct that a person may not act as a representative. (3) Any party may apply to the court for an order under paragraph (2). (4) Unless the court otherwise directs any judgment or order given in a claim in which a party is acting as a representative under this rule – (a) is binding on all persons represented in the claim; but (b) may only be enforced by or against a person who is not a party to the claim with the permission of the court”.[57]The leading authority on the application of CPR r. 19.8 is Lloyd v Google [2021] UKSC 50. I can summarise the relevant principles by reference to the authoritative examination of the rule by Lord Leggatt JSC (with whom the other JJSCs agreed) in that case, together with some additional points made in the very recent judgment of Nugee LJ in Commission Recovery Limited v Marks & Clerk LLP [2024] EWCA Civ 9 (“Commission Recovery”), as follows:a. The rule was devised in recognition of the fact that it is not always practically convenient to join all interested persons as parties to an action, particularly if they are numerous. Instead of abandoning the aim of settling the rights of all interested persons in a single proceeding, the rule allows one or more claimants or defendants to represent all others who have the same interest as them. All persons represented in this way are bound by the court’s decision. (See Lloyd v Google at [34] and Commission Recovery at [27]).b. The development of digital technologies has added to the potential for mass harm for which legal redress may be sought. In such cases “it is necessary to reconcile, on the one hand, the inconvenience or complete impracticality of litigating multiple individual claims with, on the other hand, the inconvenience or complete impracticality of making every prospective claimant (or defendant) a party to a single claim. The only practical way to ‘come at justice’ is to combine the claims in a single proceeding and allow one or more persons to represent all others who share the same interest in the outcome. When trying all the individual claims is not feasible, the adages of Lord Eldon quoted by Lord Macnaghten in Duke of Bedford v Ellis [1901] AC 1 remain as pertinent as ever: that it is better to go as far as possible towards justice than to deny it altogether and that, if you cannot realistically make everybody interested a party, you should ensure that those who are parties will ‘fairly and honestly try the right’”. (Lloyd v Google at [67]).c. Thus the representative rule should be treated as “a flexible tool of convenience in the administration of justice” and “applied to the exigencies of modern life as occasion requires” (Lloyd v Google at [68]).d. There is no limit under the rule to the number of people who may be represented. The only jurisdictional requirement is that the representative “has the same interest” in the claim as the person(s) represented (Lloyd v Google at [69] and Commission Recovery at [29]).e. The ‘same interest’ requirement must be interpreted purposively in light of the overriding objective and the rationale for the representative procedure (Lloyd v Google at [71]). In practice this means that it is enough that there is a common issue (or issues) such that the representative can be relied on to conduct the litigation in a way that will effectively promote and protect the interests of all members of the represented class. That is not possible where there is a conflict of interest between class members, i.e. where an argument which would advance the cause of some members would prejudice the position of others; but it is no impediment if the class members merely have divergent interests, i.e. where an issue may affect only some class members but advancing their case will not prejudice the position of others in the class (Lloyd v Google at [71-72] and Commission Recovery at [29]).f. It is no impediment that the members of the class all technically have separate causes of action:
“[t]he requirement is that they have the same interest in a claim in the sense of a common interest in one or more issues, not in the sense of a joint claim” (Commission Recovery at [30]). However, the claims of all members of the represented class must “stand or fall on the determination of the claims of the representatives”
. The alternative is the risk of “a form of ‘rolling’ representative action, where (at least potentially) no represented party was bound by the court’s determination of anyone else’s claim” (Jalla at [60]-[61]). g. It is a general principle that membership of the class “should not depend on the outcome of the litigation” (see Emerald Supplies Ltd v British Airways Plc [2011] Ch 345 per Mummery LJ at [62]-[63] and [65] and Lloyd v Google at [56] and [78]). However, it does not matter that the number and identities of the represented parties are difficult or impossible to ascertain, or that the class is a fluctuating one (see Duke of Bedford v Ellis [1901] AC 1 per Lord Macnaghten at page 11 and Lloyd v Google at [78]). h. Where the same interest requirement is satisfied, the court has a discretion whether to allow a claim to proceed as a representative action. In exercising its discretion, the court must give effect to the overriding objective of dealing with cases justly and at proportionate cost (see CPR r.1.1(2)). Many of the considerations specifically included in that objective “are likely to militate in favour of allowing a claim, where practicable, to be continued as a representative action rather than leaving members of the class to pursue claims individually” (Lloyd v Google at [75]). i. Relevant to the exercise of the court’s discretion is the adequacy of the definition of the class of persons represented (Lloyd v Google at [78]). j. It is primarily a matter for the claimant who wishes to act as a representative to identify the class which he wishes to represent (see Commission Recovery at [41]). k. There is no requirement for consent, only community of interest. Thus “there is ordinarily no need for a member of the represented class to take any positive step, or even to be aware of the existence of the action, in order to be bound by the result” (Lloyd v Google at [77]). l. The representative action finds its simplest application in claims for declaratory relief. All that is needed is for the declaration to be worded so that it applies not just to the representative claimant but to all those represented (Commission Recovery at [31]). m. The fact that the relief sought is monetary is “not a bar” to a representative claim but what limits the scope for claiming damages in such an action is that the assessment of damages will ordinarily depend on the position of each individual claimant, which is something that usually necessitates an individualised assessment raising no common issue and requires participation of the individuals concerned. A representative action is not a suitable vehicle for such an exercise. (See Lloyd v Google at [80] and Commission Recovery at [32]). n. In cases where damages would require individual assessment, there may nevertheless be advantages in terms of justice and efficiency in adopting a bifurcated process – as was done in Prudential Assurance Co Ltd v Newman Industries Ltd [1981] Ch 229 – whereby common issues of law or fact are decided through a representative claim, leaving any issues which require individual determination (whether they relate to liability or the amount of damages) to be dealt with at a subsequent stage of the proceedings. (See Lloyd v Google at [47]-[48] and [81]). At [84] Lord Leggatt observes that he could see no legitimate objection to a bifurcated process in Lloyd v Google itself. o. Although claims for damages normally require an individualised assessment, and hence a bifurcated process, there may be cases in which that is not necessary – for example where the entitlement to damages can be calculated on a basis that is common to all the members of the class (see Lloyd v Google at [82] and Commission Recovery at [36]).[58]In Commission Recovery, the Judge at first instance refused to strike out a representative claim brought on behalf of the clients of patent attorneys in respect of undisclosed commissions for referrals to a service provider. The Court of Appeal upheld his decision, and, having set out in detail the key principles articulated by Lord Leggatt in Lloyd v Google, Nugee LJ went on to apply those principles to the facts of the case before him. When dealing with the question of whether one or more issues common to all members of the identified class arose, Nugee LJ sought to identify what he described as the claimant’s “core proposition”, by which I understand him to mean the essence of the case that was being advanced, or, put another way, what the claimant needed to prove in order to establish liability. As described by Nugee LJ, if it was to establish its case in bribery and/or breach of fiduciary duty, the claimant needed to prove that, subject to two potential defences, it contracted with the defendant on its standard terms of business and that there was a payment of commission.[59]Nugee LJ went on to ask himself whether this was an issue which arose across the class (defined essentially as all current and former clients of the defendant that had a direct contractual relationship with the defendant on its standard terms of business from time to time) and in which members of the class had the same interest. He answered this question in the affirmative, saying at [51] “[i]f CRL is right and establishes its core proposition, I see no difficulty in the Court so declaring, and it seems to me self-evident that such a declaration would be equally beneficial to every member of the class”. He went on to explain at [52] that it did not matter that, even if the claimant succeeded in obtaining such a declaration, it would not resolve all the issues in the case, even on liability. These other issues (including disclosure, informed consent and limitation, as well as relief) might well require an individualised assessment, but it was “not an impediment to the use of a representative action that not all issues can be resolved on a class basis”. The fact that issues would remain which might require individual determination (including issues of liability and quantum) was “not a bar to resolving common issues through a representative claim”.[60]Furthermore, Nugee LJ expressed the view that “it does not matter how many members of the class will ultimately benefit from a declaration” (see [54]) and that it does not matter that the claimant might be wrong in its core proposition (see [55]). In addition, Nugee LJ rejected the suggestion that there was anything wrong in “salami slicing” liability so as to enable common issues to be determined on a representative basis even if they do not lead to a conclusion on liability. In addition to the core proposition, Nugee LJ also noted that there seemed to him to be no objection in principle to deciding at a representative stage “the common issue whether, as a matter of principle, the precise circumstances of [the defendant’s] retainer, or the client’s knowledge of market practice, are available as an answer to the claim, even though (if the answer were Yes) the question of liability to any particular client would remain to be resolved at a subsequent stage”.[61]I have looked at Commission Recovery in some detail because, during the course of submissions, I asked the Claimants to identify their “core proposition”. This prompted the production of a Note on the second day of the Hearing, in which they denied the need to establish a core proposition but, on the assumption that they were wrong about that, they identified as their core proposition: “[w]hether the copyright in an artistic or film work has been infringed by the Defendant as a result of Stable Diffusion having been trained on those works?”.[62]The Claimants submit that the level of generality at which this core proposition is formulated is acceptable because it does not give rise to a conflict of interest. Specifically they say that it does not matter that the acts of infringement alleged are different and encompass a number of individual claims, because in advancing any one of these claims the Sixth Claimant will not prejudice the position of those advancing the other claims. The class definition would be as pleaded in the Further Information.[63]Alternatively, the Claimants submit that if a greater level of granularity is required, then their alternative core propositions would be:a. Whether the Defendant has committed acts of secondary infringement of copyright by importing Stable Diffusion into the UK, otherwise than for private and domestic use, and/or possessing Stable Diffusion in the course of its business, or selling or letting it for hire, or offering or exposing it for sale or hire, in circumstances where artistic and/or film works were downloaded, stored or copied during the training and development of Stable Diffusion;b. Whether the Defendant has committed acts of primary infringement of copyright by downloading, storing or copying artistic and/or film works in the UK during the training and development of Stable Diffusion; andc. Whether the Defendant has committed acts of primary infringement of copyright by authorising the reproduction and/or communication to the public of a substantial part of artistic and/or film works by users of the image-to-image feature of Stable Diffusion in the UK.[64]If this greater level of granularity is required, the Claimants concede that the text prompts claim cannot proceed by way of a representative claim. Instead, say the Claimants, permission should be given to the Claimants to proceed with that claim without joinder pursuant to CPR 19.3; alternatively, the owners of the copyright in the six exclusively licensed works could be joined as claimants or defendants. I infer, for reasons to which I shall return, that this concession also applies to the image plus text prompts claim.[65]For present purposes I observe that I do not consider that Nugee LJ was intending to elevate his search for a “core proposition” to a legal requirement or that he was seeking to add an additional element to the test of “same interest”. Instead, it seems to me that he was merely adopting the concept of a “core proposition” as a means of focusing on whether the case raised core issues which were capable of being common issues which could be tried on a representative basis. In my judgment this is a helpful way of thinking about what the common issues might be and whether their determination is likely to benefit the entirety of the identified class.

Application of the principles to this case

[66]Taking my cue from the approach of Nugee LJ in Commission Recovery at [39]-[40], the court must consider two questions:(i) is this case within the rule at all? And(ii) if so, should the court nevertheless in its discretion direct that the Sixth Claimant may not act as representative? The first question depends upon two points; do the claims of each member of the class raise a common issue or issues and is there any relevant conflict of interest between them?[67]To this I would add that it seems to me to be necessary (certainly in this case) to consider at an early stage whether the members of the class are capable of being satisfactorily identified at all. It is common ground that it is for the Sixth Claimant to define the class which it wishes to represent (Commission Recovery at [41]). Although the adequacy of the definition of the class is a matter that goes to the court’s discretion, the question of whether members of the class have “the same interest” can only be addressed on the assumption that it is clear that any particular person qualifies for membership of the class. As Mummery LJ said in Emerald at [62], “[a]t all stages of the proceedings, and not just the date of judgment at the end, it must be possible to say of any particular person whether or not they qualify for membership of the represented class of persons by virtue of having “the same interest” as [the representative claimant]”.

Do the claims of the members of the class raise a common issue or issues?

[68]By way of reminder, Getty’s proposed class definition in its Further Information is as follows: “…the class of persons represented by the Sixth Claimant are those who are owners of the copyright subsisting in artistic works and film works that have been licensed on an exclusive basis to the First Claimant, the copyright in which has been infringed by the Defendant. The persons that fall within such class can be identified on the basis that(i) they have entered into an exclusive licence with the First Claimant in respect of artistic works and/or film works; and(ii) the exclusively licensed works include works which were used to train Stable Diffusion…”.[69]On the first day of the Hearing, the Claimants provided a supplemental skeleton addressing various criticisms made by the Defendant as to this class definition and suggesting an amendment to (i) above to read “they have entered into an exclusive licence with the First Claimant in respect of artistic works and of film works in which they own copyright”. I note that this wording in fact ties in with the relief sought by the Sixth Claimant on behalf of the Represented Parties as set out in the Further Information.[70]As a yet further fall-back position, in their skeleton for the Resumed Hearing, the Claimants proposed that they at least be permitted “to take representative proceedings on the basis of specific agreements”. Accordingly they proposed that the class definition should cover all exclusive licensors under Contributor Agreements #14 to 19 on the basis that they are in “essentially the same form as Contributor Agreement #17”. However, they observed that the Defendant’s apparent acceptance in submissions at the Hearing of the possibility of representative proceedings brought on behalf of exclusive licensors under a specific agreement demonstrates that there is nothing intrinsically wrong with representative proceedings in this case and that “in reality all agreements should be included in the class definition, which is effectively the Claimants’ primary position”.[71]The Defendant’s primary objection to the class definition originally proposed by the Claimants in their Further Information is that it is dependent upon a disputed issue in the litigation, namely whether copyright in the exclusively licensed Copyright Works has been infringed by the Defendant. The Defendant contends that this is an impermissibly circular definition and that a judgment on liability would have to be obtained before it is known whether the interests of the persons whom the Sixth Claimant seeks to represent are the same (see Emerald Supplies at [65]). The Defendant denies that the amendment suggested on the first day of the Hearing to add some additional wording resolves this problem.[72]By way of response, the Claimants say that the fact that there may be a dispute as to whether a particular individual meets the criteria for membership does not mean that the definition of the class suffers from the same objection as applied in Emerald. They emphasise that their proposed class definition does not depend upon the outcome of the litigation but is premised upon “objective facts”, i.e. is there an exclusive licence and was the Copyright Work used for the training of Stable Diffusion? Further, they pray in aid Duke of Bedford v Ellis [1901] AC 1, for the proposition that merely because the class definition includes “an element of the cause of action does not mean that the class definition does not work”.[73]In my judgment, there can be little doubt that the first sentence of the class definition identified in the Further Information is entirely dependent upon the outcome of the proceedings – the question of whether copyright has been infringed by the Defendant being a question that can only be determined at trial. Thus, if the class is dependent upon this definition, it appears to me plainly to fall foul of the principle in Emerald.[74]I am not persuaded that the class definition in Duke of Bedford v Ellis assists the Claimants in this regard. In that case, the class was “growers of fruit, flowers, vegetables, roots or herbs within the meaning of the Covent Garden Market Act 1828”. The question of whether individuals falling within this class had certain statutory rights was a matter which could only be determined at trial (the common issue in that case being whether growers within the meaning of the 1828 Act had those rights), but that did not mean that the class could not be identified at an earlier time, even though the composition of the class was subject to fluctuation – the composition of the relevant class was independent of the outcome of the action.[75]In the present case, however, the question of whether Copyright Works have been infringed cannot define the class because inclusion in the class would obviously depend on the outcome of the action. In that sense the proposal in the first sentence of the class definition is analogous to that in Emerald where the proposed representative claim was to be on behalf of all direct and indirect purchasers of air freight services affected by the alleged price fixing, i.e. dependent upon the success of the action. As Mr Bond correctly observes in his statement, “[f]ollowing a decision on the representative claim, to work out whether or not an individual or entity is part of the represented class (and therefore has the benefit of the judgment), it would be necessary to work out on an individual basis if their relevant right had been infringed. Conversely, if the claim fails and is dismissed, it would not be possible to work out which parties the judgment is binding on for res judicata purposes: the represented class would by definition have no members”. A similar point was made in Lloyd v Google at [56] and in Commission Recovery at [64].[76]However, this difficulty with the class definition appears to have been recognised by the Claimants, because in Rose 9, Mr Rose disavows any intention that membership of the class should depend upon a finding of infringement by the court. Instead he focuses on the second sentence of the class definition identified in the Further Information. As I understood Ms Lane’s submissions at the Hearing, this was also her focus. Specifically she emphasised that(i) the question of whether persons have entered into an exclusive licence with the First Claimant in respect of artistic works or film works and(ii) the question of whether those works have been used for the training of Stable Diffusion, are both matters of objective fact, capable of ascertainment in advance of trial.[77]Although I have no difficulty in accepting that the question of whether an individual has entered into an exclusive licence with the First Claimant in respect of an artistic work is quite obviously a question which is independent of the outcome of the trial (and I did not understand the Defendant to suggest otherwise), I find the second aspect of this proposed definition more difficult. As Rose 9 explains, there is no definitive list of the Copyright Works that have been used to train Stable Diffusion and the Defendant has only made limited admissions in its Amended Defence to the effect that “at least some” images from the Getty Images Websites and/or some Copyright Works were used in training Stable Diffusion and that “at least some” LAION-Subsets contain URLs referencing images on the Getty Images Websites.[78]Indeed Bond 6 makes clear that the Defendant’s admissions “do not extend to admissions that one or more of the works of each of the 50,000 (of unknown identity) authors that have licensed content to the First Claimant has been used to train Stable Diffusion”. Mr Bond goes on to explain that different filters were applied to produce each of the LAION-Subsets created during the training of Stable Diffusion. Thus, he says that “whether or not a given work was actually used to train Stable Diffusion would be wholly dependent on its own facts”. This is consistent with the Amended Defence which pleads (at paragraph 64.5) that “[t]he particular images used will depend on the starting dataset and the filters applied to it for each training run”. Mr Bond says that, in the circumstances “[t]he Claimants would need to establish that one or more works of each author were in fact(i) downloaded;(ii) by the Defendant; and(iii) in the UK”. He observes that “[t]his is not something which can be established for a whole class of licensors and a myriad of works by considering the works of just one allegedly representative…licensor”.[79]To my mind this evidence is persuasive. There seems to be no way at present to identify the members of the class and Mr Rose himself acknowledges that the enormous number of images in the Training Dataset(s), as well as the enormous number of Copyright Works in issue, mean that in practice the exercise of trying to determine which works within the Training Dataset(s) are Copyright Works “would be wholly disproportionate and practically impossible without significant resources”.[80]As Mr Rose also points out, however, the subset of Copyright Works used to train Stable Diffusion is a matter within the Defendant’s own knowledge and, to date, it has failed to identify how many images from the Getty Images Websites were in fact used to train Stable Diffusion and which images those were. This state of affairs has led to the Claimants raising the following proposed disclosure issues in the DRD, namely:
“8. How was Stable Diffusion trained, including…How many Visual Assets and Copyright Works were used in the training of Stable Diffusion 1.0 [and Stable Diffusion 2.0 and Stable Diffusion XL]”
. This issue (which maps on to the Claimants’ proposed issues for trial and is disputed) was on the list of matters also to be dealt with, time permitting, at the hearing of the Application. As drafted, it does not, of course, address the question of which Copyright Works were in fact used to train Stable Diffusion. However, in their skeleton argument for the Resumed Hearing, the Claimants asserted that if the court were to accept the Defendant’s complaint that it is not possible to identify people whose exclusively licensed Copyright Works were used to train Stable Diffusion, then “the Claimants would need disclosure not only of how many works were used in training, but which ones” (emphasis added).[81]Following a suggestion from the parties that this might be something on which they could liaise further in correspondence with a view to agreeing a way forward, I made an order at the Resumed Hearing to that effect. The upshot, though, is that there is no consensus as to how (if at all) the question of which Copyright Works have been used to train Stable Diffusion should be resolved, whether it can be done by way of sample, how that might play into the representative claim, whether it could be addressed at the First Trial, or anything else. The Defendant has accepted in its Defence that the questions of authorship, subsistence and infringement may be resolved by reference to the Sample Works (and this certainly provides a means of determining some of the key issues between the parties, albeit apparently not in relation to the image prompts claim which does not seek to rely on the Sample Works, even as representative examples), but it has expressly made no wider admissions as to any other Copyright Works owned by exclusive licensors and there are at present no proposals as to how these should be dealt with.[82]Standing back, and trying to see the wood for the trees, it is plain that the question of which Copyright Works have been used for training Stable Diffusion is not a question that can currently be determined. Indeed this does not appear, at present even to be a question which the parties had anticipated would need to be resolved at the First Trial, albeit the Claimants belatedly appear to acknowledge its significance in connection with the proposed representative claim. It is also plain that the class that is intended to be defined by reference to this question cannot be identified. The Order of 6 December 2024 requires the parties to liaise on the disclosure issue, but, if the Defendant maintains its position, then an individualised assessment of the use, if any, made by the Defendant of each exclusive licensors’ Copyright Works in training Stable Diffusion will be required simply to determine whether they fall within the class and thus have a claim for infringement. While it is possible that this issue could be dealt with by way of sampling and extrapolation, there is as yet no suggestion as to how this might be done.[83]In the circumstances I can see no basis on which the court can be satisfied that any particular person qualifies as a member of the class proposed or that it therefore has jurisdiction to permit this representative claim.[84]Even if I am wrong about this from a jurisdictional perspective, I am not persuaded in the exercise of my discretion that the proposed representative claim should be permitted to proceed to trial – certainly not on the information I have to date. The adequacy of the definition also goes to the exercise of my discretion in deciding whether it is just and convenient to allow the claim to be continued on a representative basis (see Lloyd v Google at [78]), and for the reasons I have set out above, there are serious problems around the definition of the class which play directly into the management of this case to trial. The parties are still discussing how the court should be invited to deal with the issue of which Copyright Works were used to train Stable Diffusion. As yet I have no proposals on that front, no idea as to whether any proposals that might be advanced are likely to be agreed and no clear understanding of how any proposals that are agreed (or subsequently approved by way of court order) are likely to impact on the trial. Furthermore, Rose 9 confirms that in circumstances where the Claimants do not know which Copyright Works were used and/or reproduced in the UK during the training and development of Stable Diffusion they are “unable to finalise a proposal for how the issue of subsistence and ownership in relation to the Training and Development claim should be tried”.[85]It would not be consistent with the requirements of the overriding objective to permit a representative claim in the absence of clear proposals (or indeed a clear understanding) as to how it will be dealt with at trial, whether samples will be used and extrapolated across the class and whether, and if so, how, any individualised assessments that may be required are to be bifurcated. Whilst I appreciate that in many cases the considerations included in the overriding objective are likely to militate in favour of allowing a representative claim, I cannot see how that can be the case where the court remains essentially in the dark as to what exactly is proposed, including as to the trial itself (an issue that comes into sharp focus given the proximity to the First Trial). Equally I cannot see that the court can make a reasoned assessment of the question of whether allowing a representative action to go forward would promote access to justice unless it has a better understanding of what is being proposed. I simply have no means at present of balancing the relevant competing interests.[86]As things stand I certainly cannot be satisfied that the representative claim will remove the need for an expensive and time consuming individualised assessment of numerous issues of liability and quantum relating to the proposed Represented Parties, or that this will not also create a very significant case management burden for the court. In this context I note that during her submissions Ms Lane said this:
“following the trial on liability you would have to work out potentially the membership of the class, the question of ownership and then would proceed to determine the question of relief and have the damages inquiry and so on” – an observation which rather makes my point. Of course, one way of addressing this issue might be by way of an appropriate sampling regime, together with cooperation between the parties over how any such regime could be extrapolated across the universe of potential claimants (or an appropriate order from the court). Indeed, Ms Lane immediately went on to say that even though “there would be more going on after the trial…you would still expect that that would happen by way of sample”
. However, there are of course no concrete proposals as to this at present and thus no means of knowing whether extrapolation across a particular class is likely to be viable.[87]In their skeleton argument for the Resumed Hearing, the Claimants accepted that where the Defendant plainly intends to run arguments on the interpretation of the various exclusive licences (including as to foreign law) the question of permission to take that course “cannot be decided by the court in the absence of draft amended pleadings”, just as until the position on the pleadings is decided, “there can be no decision on whether the arguments on exclusive licences (if permitted to be made) should be decided by reference to a sample and, if so, what sample”. But the fact that the complexities of this case quite obviously call for pragmatic solutions in the form of sampling, as the Claimants contend, is not a reason to permit a representative claim where it is entirely unclear at present how that sampling is to proceed and (more importantly) whether it is even possible.[88]Strictly, this deals with the Application and there is no need for me to address any of the remaining arguments advanced on each side in relation to it. The proposed and pleaded class definition does not identify a class with common interests. For the sake of completeness, however, I should address two additional points arising in argument: first an argument from the Defendant to the effect that different issues arise in respect of each Copyright Infringement Claim such that, if a representative claim were to be considered, it is necessary to understand the issues arising in respect of each of those claims; and second the Claimants’ belated proposal for a different class definition.[89]As to the former, I observe that, to my mind, the exercise of inviting the Claimants to identify their core propositions has exposed the extent to which a proposition at the very high level of generality proposed by the Claimants is incapable of satisfying the requirement that all members of the proposed class have the same interest.[90]A declaration of copyright infringement in relation to, say, the text prompts claim alone would not be equally beneficial to every member of the class because that claim is confined specifically to a limited number of examples authored by a limited number of individuals (not including the Sixth Claimant). The allegation (made in the text prompts claim) that an output from Stable Diffusion infringes copyright in a specifically identified Getty Copyright Work requires a comparison of the output obtained to the Copyright Work to determine whether the former reproduces a substantial part of the latter. This is quite obviously an individualised case-by-case assessment, as appears to have been recognised by the Claimants who have now identified only 17 Copyright Works on which they rely for the purposes of this assessment.[91]It is not Getty’s case that the use of a particular Getty Copyright Work to train one or more Stable Diffusion models necessarily gives rise to outputs from those models (when a text prompt is used) that resemble and therefore infringe that Copyright Work. Importantly, as the Claimants appear now to accept, when considered on its own merits rather than by reference to a generic proposition, the text prompts claim is not a claim that could be advanced as a representative claim by the Sixth Claimant owing to the fact that the Sixth Claimant is not said to be the exclusive licensor of any of the 17 Copyright Works on which the Claimants rely in the Amended SOCI. There is no suggestion that the Sixth Claimant even has a text prompts claim and there cannot be any issue arising in that claim in respect of which it has the same interest as the rightsholders who do have such claims. There can certainly be no suggestion that the claims of all members of the represented class can “stand or fall on the determination of the claims of the representatives” (see Jalla v Shell International Trading and Shipping Co Ltd [2021] EWCA Civ 1389, per Coulson LJ at [60]).[92]What is key is that the interests of the proposed Represented Parties are common. As the Claimants now accept, there is no core proposition that can be identified (at this level of granularity) in relation to the text prompts claim that could be pursued by the Sixth Claimant as representative or that would give rise to a declaration which would be equally beneficial to every member of the proposed class. If there is no benefit to all, there cannot be the same interest in the claim.[93]In the circumstances, I do not consider the Claimants’ approach of relying only upon a very high level proposition is appropriate. Such an approach might be said to mask the scope for argument over the lack of any real common interest, alternatively the existence of conflict. In my judgment it is important to focus at a more granular level (as Nugee LJ did in Commission Recovery) on the issues arising in respect of the pleaded claims before a decision can be made on whether those claims are capable of being brought as representative claims. That is not to say that represented parties must have the same claims, but they must, as is confirmed by the authorities, have a common interest in issues that arise in the claims. Here there is quite simply no common interest in any issue which arises in the text prompts claim.[94]Although not addressed by the Claimants in their Note on Core Propositions, I infer that their concession in relation to the text prompts claim must apply equally to the image plus text prompts claim. Looking closely at the pleaded case, the new RRAPoC pleads that “[i]n the case of an image entered by a user with a text command, in the circumstances identified in paragraphs 21A, 21B and 22 to 24 of Section C of the SOCI, the synthetic image output comprises the whole or a substantial part of one or more of the Copyright Works and/or Visual Assets” (emphasis added). Paragraph 21A of the SOCI pleads that the Claimants “rely upon the infringing Outputs attached at Annex F hereto”. These infringing outputs (of which there are various examples) relate only to five specific Copyright Works, A2, A3, A8, A14 and A16. The image plus text prompts claim thus appears to be restricted solely to these examples, none of which is authored by the Sixth Claimant. Just as the text prompts claim could not possibly proceed by way of a representative claim, in my judgment and for the same reasons, the image plus text prompts claim also could not proceed by way of a representative claim. The Sixth Claimant does not have an image plus text prompts claim and there is no core proposition that can be identified in relation to the image plus text prompts claim that would give rise to a declaration which would be equally beneficial to every member of the proposed class.[95]Finally, I have considered whether I should deal with the alternative proposal made by the Claimants as to the definition of the class by reference to exclusive licensors who had signed up to Contributor Agreements #14 to 19. Although this is a new case, not previously advanced, I recognise that it is in many ways a proposal which aligns with the first limb of the Claimants’ pleaded class definition, which I have dealt with in detail above. It is certainly a proposal that, in my judgment, ought to involve no difficulties over the identification of the class. In oral submissions, Ms Lane explained that it was put forward on the basis that Contributor Agreements #14 to 19 are in essentially the same form as Contributor Agreement #17 and that Contributor Agreements #14 to 19 were all the agreements that were “currently pleaded”.[96]I have wavered over this point but in the end I have decided that where this proposal was only raised at the Resumed Hearing without notice to the Defendant, it is neither fair nor consistent with the overriding objective to permit it. I reject the suggestion that the Defendant itself argued at the Hearing that this would be an acceptable class definition for representative proceedings; rather its point was that (until the Resumed Hearing) the Claimants had not even attempted to formulate a class definition along these lines.[97]I recognise that it is possible that a representative claim (at least in relation to the image prompts claim, the Training and Development Claim and the Secondary Infringement Claim) might be capable of being pursued on the basis of a class of exclusive licensors under specifically identified agreements. However, that is not the way that the Claimant has currently advanced its case on the pleadings and it is not consistent with the Claimants’ evidence. The Defendant is entitled to know the case that it must meet in pursuit of its Application and should not be presented with a moving target. That is particularly so where the Claimants have the carriage of the representative claim and where the responsibility for formulating that claim, including the identification of the class definition, rests solely with them.[98]Further, and decisively, it appears to me that amending the class definition in this way (even if such late amendment creates, in reality, no prejudice to the Defendant) will not improve the very serious uncertainties surrounding the management of the case that I have identified above. As things stand, the lack of clarity around so many of the central issues in this case, combined with the lack of any satisfactory proposals as to how a representative action would proceed and the swiftly advancing trial date militate strongly against an order for a representative action. Accordingly, even assuming a commonality of interests, my ultimate decision on the exercise of my discretion in relation to this proposed alternative class definition would be no different from my decision on the pleaded class definition and representative claim.

CPR 19.3

[99]Where I have refused permission for a representative claim, should I nevertheless accede to the Claimants’ alternative proposal (in respect of which they have made no formal application) that the court permit them to pursue their claims in the absence of joinder of owners of Copyright Works with whom they have concurrent rights of action, pursuant to CPR 19.3 and section 102(1) CDPA?[100]I have already set out the text of these provisions above and there is very little in the way of notes in the White Book to provide assistance on the circumstances in which the court might exercise its discretion to take this course. Indeed there seem to be very few decided cases on the point.[101]The Claimants drew my attention to Noel Redding Estate Ltd v Sony Music Entertainment UK Ltd [2024] EWHC 128 (Ch) in which Michael Green J refused to strike out a claim for infringement of copyright in sound recordings notwithstanding that (amongst other things) the claimants had failed to join all owners of the copyright. There was no proposal in the case that the claim should proceed as a representative claim. At [33] the Judge observed that “the rule is directed at protecting defendants from being subject to subsequent claims for the same relief” and he went on to say at [38] that in the circumstances of that case it was appropriate to see whether all the owners of the copyright wished to be joined, but, in the event that they did not so wish, there would be no prejudice to the Defendant.[102]The Claimants contend that if this court refuses to permit a representative claim then the interests of justice can only be served by the grant of permission under CPR r.19.3(1). However, there is nothing in their evidence to support this proposition beyond an indication at the end of Rose 9 that absent permission for the representative claim “the court will be invited to give permission for the action to continue under CPR 19.3(1) on the basis of proportionality”.[103]In their submissions at the original Hearing, the Claimants explained this “proportionality” point, submitting that the joinder of many more parties was only likely to place a huge burden on all existing parties and the court. They also maintained that (unlike in the Noel Redding case where it was feasible to ask the owners of the copyright whether they wished to be joined) it would not be proportionate in this case, given the enormous number of exclusive licensors, to take such a step. On the subject of prejudice to the Defendant, the Claimants submitted that any concerns on the part of the Defendant as to its exposure to future claims were met by the existence of the Right To Control Claims Clause in the Contributor Agreement, which Ms Lane submitted was “good enough” to protect the Defendant from further proceedings. The Claimants contend that the effect of this clause (which is in slightly different forms in different versions of the Contributor Agreement) is that the Represented Parties cannot themselves bring proceedings for copyright infringement relating to Content they have exclusively licensed to the First Claimant.[104]In response, the Defendant complains that this informal application has been raised very late in the day and that it provides no comfort that the Defendant will be protected from future claims by exclusive licensors. It submits that where the CPR has provided a particular mechanism for a representative claim under CPR r.19.8, it would be “an error of principle” to treat CPR r.19.3 as a safety net to deal with a case where the requirements of CPR r.19.8 have not been met.[105]I start from the proposition, which is not, I think, disputed, that the sheer number of exclusive licence holders with potential claims arising by reason of their concurrent right of action in respect of the Defendant’s alleged infringements raises very particular challenges. Rose 8 says that “it would be practically impossible, and in any event disproportionate, to join every single owner of the Copyright Works to these proceedings” and I did not understand the Defendant to suggest otherwise. In circumstances where I have refused to permit a representative claim, this quite obviously raises very real issues for the Claimants, and for the court, as to how the claims of the exclusive licensees are to be tried in the most effective and proportionate manner.[106]I reject the Defendant’s submission that a failure in relation to CPR r.19.8 necessarily precludes a party from relying upon CPR r.19.3. These are different rules with different requirements. If the Claimants can persuade me to exercise my discretion under CPR r.19.3 I fail to see that I am precluded from doing so merely because of the decision I have made on CPR r.19.8. However, I agree with Mr Saunders that it is important to bear in mind how the claim has been put to date and to look to see whether the evidence served by the Claimants supports the new proposed approach under CPR r.19.3.[107]It is true that there is presently no pleading as to the Right To Control Claims Clause in the exclusive licences (the main ground for the Claimants’ assertion that there can be no real prejudice to the Defendant by an order under CPR r.19.3). However, I do not consider this point to carry the day. I bear in mind that “pleadings are intended to aid in the just resolution of disputes, not obstruct their just resolution” (Commission Recovery at [69]). While I would expect to see a properly pleaded case in due course, the key question concerns whether there is likely to be any prejudice to the Defendant by reason of this proposal and that could have been addressed by way of evidence.[108]But, until Rose 9, there was no suggestion of any application under CPR r.19.3 and no formal application has ever been made. There is no evidence from the Claimants which expressly goes to the question of likely prejudice to the Defendant in the event of the court granting permission under CPR r.19.3. Instead, the question of prejudice has been dealt with in submissions which have developed over time, particularly as to the Right To Control Claims Clause as it appears in various agreements.[109]In my judgment, however, and given the potential ramifications for the Defendant of so many exclusive licensors subsequently “vexing” the Defendant with fresh proceedings, the court is entitled to expect proper evidence addressing this point.[110]Absent such evidence, I agree with the Defendant that this is simply not an application that the court can possibly accede to at this stage. The “touchstone” is prejudice to the Defendant by reason of the potential for further claims and the Claimants have failed to serve evidence from which the court can gain comfort on this score.[111]Although it appears to be the case that various of the disclosed licence agreements contain a Right To Control Claims Clause, the Defendant correctly points out that it is not a party to these agreements and cannot enforce them; the Claimants do not propose to seek undertakings from the exclusive licensors to the effect that they will not make claims against the Defendant. In any event the Defendant has not even seen the whole universe of licences that the Claimants contend are in the scope of these proceedings – not least because the Claimants were only required to disclose licences which were common to more than 500 Represented Parties. The agreements that have been disclosed appear to be subject to New York law or Alberta law such that there may, or may not, be issues of foreign law that apply to the enforceability of the relevant provision. Furthermore, it is now plain from a footnote to the Claimants’ skeleton for the Resumed Hearing that some licence agreements have Right To Control Claims Clauses in different forms and some licences do not have Right To Control Claims Clauses at all. Where that is the case, it is now accepted by the Claimants that “Copyright Works owned by licensors with these agreements would have to be excluded [from] the proceedings if permission were given pursuant to CPR 19.3”.[112]In all the circumstances I do not consider it to be appropriate, as things stand, to grant permission under CPR r.19.3 and section 102(1) of the CDPA 1988. I do not consider that such permission would be consistent with the requirements of the overriding objective, including dealing fairly with this case and ensuring that the parties are on an equal footing.

The Terms of the Order

[113]In light of my findings I invite the parties now to cooperate in preparing a draft Order which reflects my decisions and which dovetails with the Order of 6 December 2024. The Order should direct that the Sixth Claimant may not act as a representative (although it may of course remain in the proceedings as a claimant in its own right). The Order should also dismiss the Claimants’ informal application for permission under CPR r.19.3.[114]Although each party has already provided me with draft proposed Orders dealing with case management, on reflection I do not consider it to be appropriate to finalise an order dealing with the case management consequences of my decisions without further recourse to the parties. By way of guidance, however, I observe:[115]First, in relation to the Output Claims alleging infringement by use of text prompts and image plus text prompts, I see no reason why these should not proceed on the basis of the 17 Copyright Works (text prompts) and 5 Copyright Works (image plus text prompts) relied upon in the SOCI, subject either to the limited number of exclusive licensors of any of these works which are not owned by Getty being joined to the claim (either as claimants or non-participating defendants under section 102(2) CDPA), or to undertakings being provided from each exclusive licensor not to bring its own claim against the Defendant, in which case I shall permit an application that they need not be joined under CPR r.19.3 and section 102(1) CDPA.[116]This reflects a proposal made by the Defendant’s solicitors in a letter to the Claimants’ solicitors dated 20 November 2024 and appears to me to represent a pragmatic way of enabling the Claimants to pursue the text prompts claim and the image plus text prompts claim by reference to the limited number of Copyright Works on which they now seek to rely. I can see no reason to think that the Defendant will be in any way prejudiced in either scenario; on the contrary it will be protected from being the subject of subsequent claims by third parties for the same relief. Furthermore I cannot see that either solution places an inappropriate burden on the Claimants In their skeleton for the Resumed Hearing, the Claimants suggested that obtaining undertakings for the purposes of the Secondary Infringement Claim, the image prompts claim and the Training and Development Claim would be “entirely disproportionate” owing to the number of exclusive licensors involved. There was no similar submission in relation to the text prompts and the image plus text prompts claims. . Ultimately, however, I leave it to the Claimants to determine whether they wish to adopt either of these options. If not, then these claims must proceed as envisaged in paragraph 2 of the draft Order provided by the Defendant.[117]Second, I expect the parties to be able to agree on paragraph 1(a) of the Order in respect of the Training and Development Claim in light of the agreed position arrived at during the hearing. In light of my decisions in relation to CPR r.19.8 and CPR r.19.3, I do not consider it to be appropriate to include in the Order a paragraph in the terms proposed by the Claimants at paragraph 2 to their proposed draft Order.[118]Third, I do not presently see why the Secondary Infringement Claim may not proceed by reference both to Sample Works A to D (which are owned by the First Claimant) and Sample Works E to K. The latter have been exclusively licensed by the Sixth Claimant to the First Claimant and the Sixth Claimant remains a party to the proceedings. This has the advantage that points that the Defendant wishes to raise in relation to CA#17 (or any other licence agreements that are said to cover the Sixth Claimant) may be raised at the First Trial – always subject to those points being properly pleaded.

Conclusion

[119]Notwithstanding that the Claimants have ultimately been unsuccessful in convincing the court to permit the representative claim, it is abundantly clear that the difficulties I have identified are not intractable and that a pragmatic way forward must be found which does not involve the joinder of 50,000 potential claimants. It is a function of the way in which the CPR r.19.3 application has been pursued that I have determined it to be unfair to make the order sought. However, that does not mean that such an order could not be obtained upon the service of appropriate evidence and I have no intention of shutting the Claimants out from applying again, if they wish to do so.[120]Indeed, in many ways an order under CPR 19.3 would make very good sense. With such an order there would be no need to identify common issues or to consider bifurcating the trial. The issues arising on the claim could potentially be dealt with by sample (agreed by the parties or imposed by the court) and the focus could be on the “big picture” issues. There would be no need (as Mr Saunders accepts) to open up what he describes as “Pandora’s box” of exclusive licences – i.e. the true interpretation of their individual terms. There would also be no need for anyone other than the First Claimant to pursue the claims; the Sixth Claimant would not need to be involved, because his sole role is as representative of the Represented Parties. It is not the Claimants’ case, as I understand it, that it is necessary to introduce all the complexity potentially associated with the exclusively licensed works in order fairly to determine the core issues in this case at trial.[121]The Claimants contend in their skeleton for the Resumed Hearing that “it is clear that Getty Images has the exclusive right to control proceedings related to the licensed works”. Moreover that “there is no provision which allows the licensor to bring proceedings if Getty Images has elected not to do so”. No doubt by way of testing this proposition, the Defendant suggested that the Claimants might provide an undertaking to indemnify the Defendant against all damages, legal costs and other expenses that it might incur as a result of a subsequent claim brought by an exclusive licensee. Ms Lane’s response to this suggestion was that this had been considered but “the clauses are themselves so clear” that an undertaking was “not necessary”. The Claimants subsequently confirmed after the Resumed Hearing that they were not prepared to provide undertakings of the sort suggested.[122]It is not for the court to tell the Claimants how they should proceed, but I merely observe that at the heart of this debate lies a very simple issue as to the apportionment of risk. If they are to convince the court to make a CPR r.19.3 order the Claimants will need to satisfy it that there is no risk to the Defendant of future claims. If, as they say, there really is no risk of future claims because the wording of the relevant clauses is clear, it is difficult to see what possible downside there could be to Getty taking that risk by the provision of the proposed, or a similar, undertaking – even if it be purely by reference to an identified subset of agreements. I note Mr Saunders’ indication that:
“if their case is they can do the whole thing without recourse to these parties because they are not people entitled to the remedy in themselves, following the licence agreements they have executed, then…we do not care. Off we go with Getty”
. Accordingly it would appear that, subject to appropriate comfort being provided, the Defendant would be inclined to consent to an order under CPR r.19.3.[123]An alternative approach might be for the Claimants to narrow the universe of Represented Parties in a way that allows a reasonably circumscribed and manageable set of legal determinations to read across the whole. As I understand the Defendant’s submissions, this was not, in theory, a solution to which they would necessarily be averse.[124]In any event, it is imperative that the parties now seek to resolve, as quickly as possible, the serious case management issues that have been exposed by the making of the Application and that they find a way of taking this case forward to the First Trial in June on the basis of a realistic and manageable set of issues. I have already determined that further case management conferences are to take place in January and March 2025 with a view to ensuring that ongoing, tight, case management is possible. At the very least I shall expect by then to see final proposals from both sides for the management of the case to trial, together with a list of issues for trial.

Cited in 1 later judgment