Shaun Thompson & Anor, R (on the application of) v The Commissioner of Police of the Metropolis

This claim challenges the legality of the policy of the defendant, the Commissioner of Police of the Metropolis, adopted on 11 September 2024 in relation to the deployment in London of live facial recognition technology (“LFR”). The policy is entitled “MPS Overt LFR Policy Document” (“the Policy”).

LFR is seen by the defendant and other police forces as a powerful tool for the prevention and detection of crime, finding missing persons and the protection of the public. It involves the use of cameras, which are generally moveable rather than fixed, to capture the images of members of the public in a public location. The facial features of individuals are expressed as unique biometric or numerical values. The software then compares those values with the biometric details of a number of persons sought by the police to see whether they achieve a pre-set score indicating the required likelihood of a positive match.

If there is no match, the biometric data of the member of the public is automatically and immediately deleted. Any facial images detected by the system which have not been matched are blurred in the images visible to officers. But if the biometrics do match, an alert is created and the two images to which the data relates are examined by officers to decide what if any action to take. The technology is capable of scanning a substantial number of faces per second and a very large number over the course of a deployment.

The claimants do not contend that the use of LFR is unlawful as a matter of principle, or in all circumstances. However, they submit that the use of LFR gives rise to “significant civil liberty concerns” which have become greater with the increased deployment of LFR over the last few years. They rely upon a number of matters. LFR involves the extraction of a person’s biometric data, a key attribute of an individual given its unique characteristics. LFR can be deployed without the consent or knowledge of the persons involved, on a large scale or in crowded locations. The overwhelming majority of persons whose facial biometrics are captured and processed are not suspected of any wrongdoing and are not otherwise persons for whom the police are looking. There is a risk of false identification which may result in intrusive questioning to establish identity.

However, it is important to emphasise that neither this case, nor this court, is concerned with the merits and demerits of the use by the Metropolitan Police Service (“MPS”), or other police forces, of LFR. Judicial review is simply the means of ensuring that public bodies act within the limits of their legal powers, and in accordance with the Human Rights Act 1998, as well as any relevant procedures and legal principles governing the exercise of their decision-making functions. The role of the court is concerned with resolving questions of law. It is not responsible for making political, social or economic choices, or in this case policy decisions, on how the police carry out their functions. Such decisions are ones that Parliament, or the common law, has entrusted to the relevant public authorities, not the court. The court is only concerned with the legal issues which the claimant is entitled to raise in these proceedings.

The claimants contend firstly, that the Policy violates Article 8 of the European Convention of Human Rights (“ECHR”), in that, contrary to Article 8(2), it allows “interference” with the “right to respect for private and family life” by the use of LFR which is not “in accordance with the law” (“IAWL”). This was the first ground of challenge argued at the hearing before us (“ground 1”).

Secondly, the claimants contend that the Policy violates Articles 10 and 11 of the ECHR in that, contrary to Article 10(2) and Article 11(2), it allows “restrictions” upon the “freedom of expression” and the “freedom of assembly and association” by the use of LFR which is not “prescribed by law” (“PBL”). This was the second ground argued at the hearing before us (“ground 2”).

It is common ground between the parties that in the present case whether the Policy is IAWL raises the same issues as whether it is also PBL. The court’s conclusion on the IAWL issue, either way, will also hold good for the PBL issue.

Mr Dan Squires KC who, together with Mr Aidan Wills and Ms Rosalind Comyn, appeared for the claimants, stated that grounds 1 and 2 advance essentially the same contentions in relation to Articles 8, 10 and 11, namely whether the Policy leaves too much discretion to individual officers within the MPS to have the quality of “law.” In that regard, the claimants accept that the defendant has the legal power to use LFR and to adopt a policy for its deployment.

We should at this stage note that some of the authorities cited to us use the expression “the legality principle” to refer to “the quality of law” requirement. Irrespective of nomenclature, we are concerned with the principle derived from the caselaw of the ECHR, and applied in the domestic cases, that a measure, such as the Policy, must have the “quality of law” in the sense that it must be accessible to the persons concerned, foreseeable as to its consequences and compatible with the rule of law (Huvig v France (1990) 12 EHRR 528 and Kruslin v France (1990) 12 EHRR 547, applied by the Supreme Court in In Re Gallagher [2019] UKSC 3; [2020] AC 185 at [14]).

There is no issue in the present case about the accessibility of the Policy. It has been published and it is comprehensible. The contrary has not been suggested.

The central issue in this case relates to foreseeability. Mr Squires frequently submitted that the Policy (or rather certain parts of the Policy) confers too broad a discretion on individual police officers when deciding on whether and where to deploy LFR. But this turned out to be a form of shorthand. He accepted that the appropriate test is that stated by Lord Sumption JSC in Gallagher at [17]:

In other words, a measure must have sufficient clarity and foreseeability for the purposes of IAWL and PBL so as not to allow arbitrariness, that is decision-making by a public authority on the basis of whim, caprice, malice or predilection (Lord Bingham in R (Gillan) v Metropolitan Police Commissioner ([2006] UKHL 12; [2006] 2 AC 307 at [34]). This court is being asked to determine whether particular parts of the Policy identified by the claimants are contrary to Articles 8, 10 and 11 because they allow officers to make individual decisions on the deployment of LFR which are arbitrary. The Divisional Court and the Court of Appeal have previously considered these issues in relation to LFR: R (Bridges) v Chief Constable of South Wales Police (Information Commissioner and others intervening) [2019] EWHC 2341 (Admin) [2020] 1 WLR 672; [2020] EWCA Civ 1058, [2020] 1 WLR 5037), which we discuss below.

It is important to be clear at the outset that a challenge to the foreseeability of a measure, or its alleged arbitrariness, does not involve a challenge to the proportionality of that measure or its application in a particular case. That distinction was emphasised by Lord Sumption in Gallagher at [14] and [17]. Lord Reed JSC had previously drawn the same distinction in R (T) v Chief Constable of Greater Manchester Police [2014] UKSC 35; [2015] AC 49 in his analysis at [114]-[115] (endorsed by Lord Sumption in Gallagher at [36]-[41]).

Whether a measure satisfies the requirement of foreseeability is an anterior legal question. If it does not, then the measure violates whichever Convention right is engaged. If it does, then issues as to justification and proportionality may fall to be determined. However, Mr Squires made the claimants’ position clear. They do not raise any issue regarding justification or proportionality in this case.

The first claimant is a 39-year-old Black man who was born and lives in London. He volunteers as a community worker with Street Fathers to provide a proactive and positive male presence for children and young people in areas of London affected by youth violence to prevent them coming into contact with the criminal justice system (Amended Statement of Facts and Grounds para.20).

On 23 February 2024 near London Bridge station Mr Thompson was stopped, detained and questioned by police officers after having been matched by LFR with a different person on a police “watchlist” (a list of “Sought Persons”). That person was his brother who was on bail for a suspected offence of inflicting grievous bodily harm on him. Although officers doubted whether the first claimant was the person on the watchlist, they nevertheless scrutinised him, asked questions and made him prove his identity. When he declined to allow his fingerprints to be taken, he was threatened with arrest. Mr Thompson explains how the experience caused him to feel angry and distressed.

The second claimant, Ms Silkie Carlo, is the director of the civil liberties organisation, Big Brother Watch. She also lives in London and has monitored and campaigned against the use of LFR by the MPS since its inception. To that end she has attended LFR deployments. In addition, Ms Carlo often attends or organises protests and is concerned about the potential use of LFR at such events, as well as the increasing use of this technology over the years.

The defendant does not dispute that both claimants have the status of victims for the purposes of their claims based on the ECHR.

The claim was originally issued on 24 May 2024. The original ground 1 alleged that the use of LFR on 23 February 2023 breached Mr Thompson’s rights under Article 8 of the ECHR on the basis that the defendant’s policy then in force breached the IAWL requirement. It was contended that that policy contained insufficient constraints on who could be placed on a LFR watchlist and where LFR could be deployed. The original ground 2 alleged that the use of LFR by the MPS breached Ms Carlo’s rights under Articles 10 and 11 of the ECHR because its then policy on deployment breached the PBL requirement.

The original claim for judicial review was accompanied by an expert report by Professor Martin Utley, a Professor of Operational Research at University College London, which sought to show that the defendant’s policy then in force allowed the deployment of LFR across large areas of London, although the policy purported to “limit the locations” where LFR could be used.

On 4 June 2024 the defendant wrote to the claimants’ solicitors seeking their agreement to a stay of the claim while he undertook a “thorough and detailed” review of the LFR policy.

The claimants did not agree to the claim being stayed and so the defendant applied for a stay until 11 September 2024, the date anticipated for the completion of the review. On 22 July 2024 Sheldon J granted the stay. He also granted the claimants permission to amend their claim in the light of the review.

As we have said, on 11 September 2024 the defendant did publish his revised and current Policy, which replaced the previous policy documents.

On 29 November 2024 the claimants filed an Amended Statement of Facts and Grounds which included challenges to the new Policy relying on an alleged failure to comply with the IAWL or PBL requirements in Articles 8, 10 and 11 of the ECHR. Those grounds 3 and 4 correspond to what were referred to as grounds 1 and 2 during the substantive hearing before us.

This amended pleading was accompanied by an application to rely upon a second expert report by Professor Utley which gave his mathematical analysis of the effect of certain criteria in the September 2024 Policy. The claimants no longer relied upon Professor Utley’s first report.

On 18 December 2024 a consent order was agreed (eventually sealed on 30 April 2025) by which grounds 1 and 2 of the original claim were withdrawn on terms as to costs and a settlement sum payable to Mr Thompson.

On 30 April 2025 Farbey J granted the claimants permission to apply for judicial review in relation to the then grounds 3 and 4 and she directed the holding of an interim applications hearing.

On 2 July 2025 Farbey J granted the Equality and Human Rights Commission (“the Commission”) permission to intervene by way of limited written submissions and reliance upon the written evidence of Ms Angharad Davies, the Commission’s project lead on compliance work relating to police use of LFR since October 2023. That witness statement was amended on 15 July 2025.

The interim applications hearing took place on 8 July 2025. On 16 September 2025 Farbey J issued her order for directions. She ordered that the second report of Professor Utley and the third witness statement of Ms Carlo would be considered by the court de bene esse. She ordered the claimants to file a list of legal propositions which they seek to base upon Professor Utley’s second report. In her judgment handed down on the same day as her order, Farbey J dealt with the evidence of Professor Utley and Ms Carlo and cost capping and she directed that the grounds referred to as 3 and 4 in the Amended Statement of Facts and Grounds be renumbered as 1 and 2 respectively.

On 22 July 2025 the claimants filed a statement summarising the evidential points in Professor Utley’s report upon which they rely and their legal contentions to which those points are said to be relevant. The claimants say (at para.7) that on their understanding of the law on IAWL and PBL “it cannot simply be left to individual officers’ discretion to use LFR at any crowded location the officer chooses” and that if the defendant “ is able … in practice, to select virtually any area of high footfall across London and deploy LFR there in the hope that someone he seeks will be present …. [his] use of LFR is not IAWL pursuant to the principles set out in Bridges.” The claimants then say that Professor Utley’s report “goes to whether the [Policy] operates in that way, and, in practice, it prevents officers from being able to select virtually any crowded location they wish in London for deployment of LFR.” Similarly, after summarising in para.8 key points in Professor Utley’s report, the document concludes at para.9:

On 10 October 2025 the defendant filed a second witness statement by Ms Lindsey Chiswick, the Director of Performance and Insight at the MPS since November 2024 and previously Director of Intelligence. She had taken over responsibility for LFR from a Deputy Assistant Commissioner in November 2019. She is also the National Police Chiefs’ Council Lead for Facial Recognition. Her witness statement responds to the evidence filed on behalf of the claimants upon which they still rely.

Subsequently, the following further written evidence has been filed:

Article 8 provides:

Article 10 provides:

Article 11 provides:

In Bridges the Court of Appeal at [55] endorsed the Divisional Court’s summary of the general principles relating to the IAWL test:

It is important to note that in endorsing principles (2) and (3), the Court of Appeal accepted that the breadth of a discretion is not a freestanding test for determining whether that power is IAWL. The key issue is whether the scope of, or the conditions for exercising, the discretion are defined sufficiently clearly so as not to authorise arbitrary decision-making.

Principle (4) in the passage cited in [37] above was not intended to elide the distinction between IAWL and proportionality. The judgment of Lord Hughes JSC in Beghal at [31]-[32] also deals with cases where discretion or flexibility is needed as a safeguard against the automatic operation of a rule. He then explained in [33]:

Having set out the general principles, we turn to the relevant elements of IAWL in further detail.

As set out in the quotation from Bridges in [37] above, the Policy must have some basis in domestic law. It is therefore appropriate to identify the source of the powers of the police to use LFR. This was addressed in Bridges by the Divisional Court at [2019] EWHC 2341 (Admin); [2020] 1 WLR 672 at [68]-[78]. There was no dispute about that analysis in the Court of Appeal ([2020] 1 WLR 5037 at [38]). Police officers owe the public a common law duty to prevent and detect crime, which includes taking all steps appearing to them necessary for those purposes (Divisional Court at [69]). That function can include the use, retention and disclosure of images of individuals ([70]).

In R (Catt) v Association of Chief Officers of England, Wales and Northern Ireland (Equality and Human Rights Commission intervening) [2015] UKSC 9; [2015] AC 1065, Lord Sumption stated at [7] that at common law the police have the power to obtain and store information for policing purposes, that is broadly speaking for the maintenance of public order and the prevention and detection of crime. But those powers do not authorise “intrusive methods” of obtaining information, such as entry onto private property or acts which would constitute an assault.

The Divisional Court in Bridges went on to hold that the police may make reasonable use of a photograph of an individual for the prevention and detection of crime, the investigation of alleged offences and the apprehension of suspects unlawfully at large; and they may do so “whether or not the photograph is of any person they seek to arrest or of a suspected accomplice or of anyone else”, so long as they make no more than reasonable use of the image for achieving those purposes [72]. Accordingly, those common law powers were sufficient, and the police did not need any additional statutory powers, for example, to use CCTV or to use body-worn video, traffic and ANPR cameras. However, specific statutory powers were needed for “intrusive methods” in the sense used by Lord Sumption in Catt, for example, to take fingerprints or a DNA swab or to obtain a warrant to enter property ([73]-[74]). The Divisional Court then stated that LFR is not an “intrusive” method in the physical sense used in Catt.

The compilation of lists with photographs of sought after persons, or watchlists, falls within the statutory powers of the police in relation to arrested persons and their common law powers in relation to other persons of “possible interest”, as steps necessary for keeping the peace, preventing crime or protecting property ([76]-[78]).

As we have indicated, the claimants in the present case have not taken issue with any part of this analysis. Instead, the case turns on whether the legal framework governing the deployment of LFR, primarily the defendant’s September 2024 Policy, meets the requirement of foreseeability, in particular that it must not permit arbitrary decision-making.

The claimants cited three decisions of the European Court of Human Rights (“ECtHR”) which are helpful on the concept of “foreseeability” as a qualitative requirement of law.

Malone v United Kingdom (1984) 7 EHRR 14 was concerned with whether the regime then in place for telephone interception in the UK violated Article 8 of the ECHR. At [67] the Grand Chamber said this:

For intrusive forms of surveillance Article 8 does not expect that the domestic law should enable a person to see how the state’s powers will be exercised in relation to him so that he can adapt his conduct accordingly. Instead, the quality of law required is that the law should be sufficiently clear that citizens have an adequate indication of the circumstances and conditions in which a power may be exercised. In other words, the scheme should be sufficiently clear as to how it may be operated so as to avoid arbitrariness.

So in Rotaru v Romania (2000) 8 BHRC 449 the Intelligence Service’s system for keeping secret files on the applicant’s personal life was not IAWL for the purposes of Article 8. The domestic law did not lay down limits on the exercise of the power, such as the types of information that could be recorded, the categories of people who could be the subject of surveillance, the circumstances in which the powers could be exercised, the procedures to be followed and the period for which information could be kept [57]. Taking into account also the absence of supervision procedures [59], the Grand Chamber concluded at [61] that:

Essentially the same approach was taken by the Grand Chamber in a case far removed from intrusive surveillance. Martinez v Spain (2015) 60 EHRR 3 was concerned with the dismissal of a religious education teacher. Under an agreement between the state and the Vatican, such teachers were to be re-appointed annually unless the relevant Bishop gave a contrary opinion. Here the Pope gave the applicant a dispensation from celibacy which involved his loss of clerical status and a prohibition on teaching the Catholic religion unless the Bishop should decide otherwise. The Bishop refused to approve the continuation of the applicant’s role as a teacher of religious education. At [117] the Grand Chamber said:

The court decided that the legal arrangements in place for dealing with personal conduct satisfied the “quality of law” test. It was sufficient that on the basis of the wording of the agreement between Spain and the Vatican, the applicant could have reasonably foreseen that in the absence of a certificate of suitability from the Bishop, his contract would not be renewed.

In Bridges the Court of Appeal accepted at [82]-[83] that it is permissible to take a “relativist approach” to compliance with the foreseeability requirement: “… the more intrusive the act complained of, the more precise and specific must be the law said to justify it”. In this context it is necessary to have in mind that the LFR regime does not involve covert surveillance, nor does it involve the regulation of a person’s conduct. Instead, it is a technique for the rapid identification of persons who are already on a police watchlist. For those who are not on a watchlist, there is no prior collection and storage of data and, where they are not matched with someone who is on a watchlist, their image is almost instantaneously deleted.

In Gillan at [34] Lord Bingham gave this pithy explanation of what is meant by arbitrariness such as to contravene the IAWL requirement:

Mr Squires repeatedly submitted that this court needs to consider whether the discretion allowed to police officers operating LFR is sufficiently circumscribed or sufficiently curtailed. If public authorities are left with an “excessively broad discretion” then the relevant measure will not be IAWL. With respect, this submission is itself expressed too broadly, as two citations will show.

The ECtHR in Gillan v United Kingdom (2010) 50 EHRR 45 disagreed with the decision of the House of Lords that stop and search powers under the Terrorism Act 2000 were IAWL for the purposes of Article 8. After setting out the components of the quality of law requirement the ECtHR said this at [77]:

In Gillan the guidance given to individual police officers on the use of stop and search focused on the manner in which it should be carried out rather than giving any criteria for, or imposing any restriction on, an officer’s decision to stop and search. That decision was based exclusively on the “hunch” or “professional intuition” of the officer concerned. There was no requirement for the officer to be able to show a reasonable suspicion, or even a subjective suspicion, of anything to justify the use of the power [83]. It is therefore understandable that the Court referred in [77] to unfettered power.

Consequently, the ECtHR concluded that the grant of such a broad discretion to a police officer, as described, carried with it a clear risk of arbitrariness [85]. In other words, the Court did not conclude that Article 8 was violated simply because the discretionary power was “insufficiently circumscribed”. Instead, the key part of the reasoning was that the unfettered or unconstrained nature of the power given to officers allowed them to make arbitrary decisions. Arbitrariness is the touchstone for deciding whether this part of the quality of law test is satisfied or breached. The same principle can be seen at work in the most recent Strasbourg decision cited to us, AR v United Kingdom Application No. 6033/19, 1 July 2025 at [60]-[68].

Gillan may be contrasted with R (Roberts) v Commissioner of Police of the Metropolis [2015] UKSC 79; [2016] 1 WLR 210, where the Supreme Court considered the more detailed constraints imposed on stop and search powers directed towards knife crime and the use of other offensive weapons. The exercise of the power had inter alia to be based upon a reasonable belief in grounds, supported by evidence, which made the use of that power necessary and proportionate to the danger contemplated. The authorisation had to cover a defined area for a limited period of time. A written record had to be kept of the grounds and material relied upon. When making a stop the officer was required inter alia to explain to the person concerned the power under which he was acting, the object of the search and why he was doing it. The Court held that the law was IAWL and compatible with Convention rights. They considered that these various requirements would make it possible to assess whether there was compliance with Article 8 in practice [42]-[48].

In In Re Gallagher Lord Sumption JSC said at [17]:

Thus, the mere fact that a discretionary power is broad is insufficient to make that power not IAWL. A key question remains whether the exercise of the power is subject to principles, whether set out in law, policy or a combination of the two, which suffice to prevent decisions being dependent on the will of the decision-maker, or arbitrary (as explained by Lord Bingham in Gillan).

The Divisional Court decided that the use of LFR does entail interference with Article 8 rights [52]-[62]. Although the mere taking of a photograph by the police of a citizen in a public place would not amount to such an interference, LFR goes further. Each image is analysed and biometric facial data extracted. That represents unique information which identifies a person. The data is then used to make a comparison with persons on the police watchlist [54] and [56]-[57]. The fact that, save where a match is detected, an individual’s facial information is retained for only a very short period before being discarded, is nevertheless sufficient to involve an interference with Article 8 [59].

The Divisional Court dealt with the IAWL issue at [63]-[97]. It first decided that the police had proper legal powers to authorise the use of LFR [68]-[78].

At this point, it is worth noting that the claimant, also represented by Mr Squires, submitted that absent certain safeguards, the use of LFR would not be IAWL (see [64]). He submitted that those safeguards were:

At [85]-[97] the Divisional Court reviewed the various components of the legal framework governing the use of LFR by South Wales Police. It comprised:

The Divisional Court concluded that taking these matters together, the interference with Article 8 rights through the use of LFR was governed by a legal framework sufficient to be IAWL. The court concluded that it was neither necessary nor practical for the framework to define precise circumstances in which LFR might be used, such as which offences could justify including a person on a watchlist or what the “sensitivity settings” should be [96].

The Divisional Court went on to reject the claimant’s contention that his Article 8 rights had been violated applying the proportionality test in Bank Mellat v HM Treasury (No.2) [2013] UKSC 39; [2014] AC 700. The court concluded inter alia that any interference with the claimant’s Article 8 rights would have been very limited. It would have been restricted to the near instantaneous algorithmic processing and discarding of his facial biometric data. No personal information would have been available to any police officer or human agent. No data would have been retained [101].

The Court of Appeal disagreed with the Divisional Court on the IAWL issue, but they agreed that there had not been any violation of the claimant’s Article 8 rights.

Although LFR may be a potentially evolving technology, the Court of Appeal stated that it was neither necessary nor helpful for the court to be asked to consider hypothetical scenarios which may arise in the future. Whether other uses by the police of LFR in the future will be lawful would be a matter to be considered at that stage if the facts of such a case should arise in practice [60]. So we decline to take up the claimant’s invitation to consider consequences which could flow in future from permanent camera systems (such as the one recently installed in part of Croydon) or the suggestion that LFR could be linked to CCTV systems. That would involve speculation.

The issue in Bridges related to the deployment of LFR in the area of one police force. It is well-established that an adequate legal framework may rely on policies which are local rather than national in coverage [61].

The LFR regime in that case involved overt, not covert, surveillance [64]. But LFR is not simply analogous to the police taking photographs of people in a public place or the use of CCTV cameras. The Court of Appeal decided at [85] that LFR “falls somewhere in between the two poles on a spectrum” represented by S and Marper v United Kingdom CE: ECHR: 2008: I204; (2008) 48 EHRR 50 and R (Catt) v Association of Chief Police Officers [2015] UKSC 9; [2015] AC 1065; (2019) 69 EHRR 7. S was concerned with the retention of fingerprints and DNA samples. Catt concerned the collection, retention and use of information on an “extremism database”, which included participation in demonstrations, some involving serious disorder and criminality, although not on the applicant’s part. However, it should be noted that in both cases the ECtHR did not rule on the IAWL issue. Instead, each claimant succeeded on proportionality grounds.

The Court of Appeal’s positioning of LFR on its spectrum involved the following factors [85]-[89]:

The Court of Appeal concluded that the legal framework applicable in South Wales lacked the quality of law needed to avoid arbitrariness. They had two areas of concern, the “who question” and the “where question”. It was not clear who could be placed on the watchlist, nor was it clear that there were any criteria for determining where LFR could be deployed. Significantly, the Court found that data protection legislation and the LFR policies merely required that deployment should be necessary for a proper law enforcement purpose [92]. In a passage emphasised by Mr Squires, those uncertainties, and even lack of criteria, left too much discretion to individual police officers in those important areas [91].

However, the Court approved of one element of the deployment of LFR in South Wales which it regarded as “crucial”, namely the automatic and almost instantaneous deletion of the data of any person not matched with someone on a watchlist, with no opportunity for human observation. The Court indicated that policy for the use of LFR should make it clear that such automatic and instantaneous deletion is required [93].

The Court then reviewed in some detail the constituent parts of the legal framework relied upon by the South Wales Police force [98]-[130]. The legal protections in the Data Protection Act 2018 formed an important part of the framework, but by themselves were insufficient to satisfy the IAWL requirement [104].

The Court found that the Surveillance Camera Code of Practice did not provide criteria for inclusion on a watchlist and what should be addressed in local policies for the location of LFR deployment [118] and [120].

In relation to the policies of the South Wales Police force, the court said at [121] that the critical question was how much discretion those policies gave to individual officers (in the sense explained in [37]-[38] above) as to who could be put on a watchlist and where LFR could be used. On the “who question” the Court decided that, in contrast to objective criteria which asked whether a person was suspected of an offence, wanted on a warrant or vulnerable, a criterion which simply referred to “other persons where intelligence is required” was too uncertain and subjective [123]-[124]. Moreover, the Standard Operating Procedure did not contain any “who” criteria [128]. Lastly, none of the documents had any “where” criteria [130].

The Court of Appeal stated that it was not for a claimant or for the court to design a particular set of policies for the police to comply with the quality of law requirement. However, the court was satisfied that the force’s policies did not sufficiently set out the terms onwhich discretionary powers can be exercised by the police and for that reason did not have the necessary quality of law [94]. The court also made it clear that did not wish to be unduly prescriptive as to the content of any new policies [95].

In reviewing the defendant’s revised Policy in September 2024 against the quality of law requirement, we will bear in mind the relativist approach approved by the Court of Appeal in Bridges and the factors which (as we have set out in [70]-[71] above) point to the use of LFR as falling somewhere in between the two poles on a spectrum which are represented by, on the one hand, the storing of photographs and intelligence notes on a database and, on the other hand, the retention of fingerprint and DNA samples.

In this context it is also relevant to have regard to some of the conclusions of the Court of Appeal when they rejected the claimants’ proportionality ground of challenge. The Court concluded that the impact of the deployment of LFR on the Article 8 rights of the claimant, who was not on the watchlist and was not stopped by the police, was negligible, as would be the case for any other member of the public in the same position. The Court added that the very little weight given to that impact did not become weightier because other people were similarly affected. This was a matter of judgment not multiplication [143].

The MPS has used LFR since 2016. After a trial phase, it became fully operational in January 2020 using technology known as the NEC-3 algorithm. The deployment of LFR was governed by an initial policy framework that was in force at the time of the false identification of Mr Thompson. We have not been directed to any part of that earlier framework.

On 14 March 2024, the defendant decided to conduct a review in order to assess whether LFR policy should be adjusted in light of operational experience. At the conclusion of the review, the defendant withdrew the initial policy documents and introduced the Policy that is the subject of the present claim.

The present Policy came into effect on 11 September 2024 and it has been published. It concerns the overt use of LFR that members of the public can know about in advance and detect as they move around London by virtue of notices on the MPS website and social media, as well as measures at the site of the LFR deployment such as MPS-liveried signage (paras 10.2-10.3 of the Policy). It does not concern the use of LFR technology for covert police operations with which this claim is not concerned.

The Policy governs the deployment of a dedicated CCTV system used to locate “Sought Persons” (para.2.1). A selection of pre-existing images of Sought Persons will be placed on one or more “watchlists.” The LFR software detects and analyses the facial features of the images in the selection and expresses those features as a set of numerical values known as a “Biometric Template” (para. 3.1(1)).

LFR cameras provide live CCTV footage of members of the public who appear within the “Zone of Recognition” (para.3.1(2)). This is the area in which a camera can generate reliable facial images and is generally smaller than a camera’s field of view (Annex A of the Policy).

The Biometric Template of anyone passing through the Zone of Recognition is compared by the LFR software with the Biometric Templates on the watchlist. For each Biometric Template created by the software, a “similarity score” is generated indicating the extent of similarity between that Template and the Templates on the watchlist. If the score exceeds a pre-set number, the LFR software will generate an alert to indicate a possible match (para.3.1(6)).

After an alert has been generated, specially trained police officers – known as Engagement Officers – will assess the images and make a decision as to whether the match is “viable” and, if so, whether any further action is required (para.3.1(7)). In the absence of an alert, any Biometric Template created by LFR is immediately and automatically deleted (para.3.1(8)).

The CCTV footage is viewed in real time by an LFR Operator for the purposes of ensuring that the LFR system is working properly and to support Engagement Officers. The images of those within the CCTV footage who do not meet the similarity score are blurred. This blurring means that no-one viewing the CCTV footage is able to see the faces of anyone who has not been the subject of an alert (para.3.1(6)).

Subject to exceptions which are irrelevant to the present claim, all CCTV footage generated by LFR technology is deleted within 31 days. Watchlists are deleted as soon as reasonably practicable and in any event within 24 hours (para.3.1(8)).

None of what we have described so far was in dispute. The parties focused their submissions on three significant sections of the Policy, namely:

We shall describe each of these aspects of the Policy in turn. But we should note para.1.3 which states that:

It is plain, and not in dispute, that these limbs of para.1.3 are cumulative. The provisions to which they refer form part of the interlocking requirements of the Policy.

Under section 2 of the Policy, the purpose of LFR is (as we have already said) to locate “Sought Persons” whose names have been placed on one or more “watchlists.” In addition, LFR may “serve to achieve the MPS’s policing objectives by helping to deter or disrupt criminality or other conduct posing a risk to public safety” (para.2.1).

Despite this broad purpose, the Policy contains significant constraints because LFR technology may only be used in three specific sets of circumstance, known respectively as Use Cases A, B and C (paras 2.2-2.12). We describe each of these Use Cases below, while noting that in his oral submissions on this issue Mr Squires focused solely on “crime hotspots” which form part of Use Case A.

Under Use Case A, the MPS may deploy LFR to support the policing of “hotspots” (para.2.3). There are two kinds of hotspot: a “crime hotspot” and a “missing person hotspot.”

Under paragraph 2.3(a) of the Policy, a crime hotspot is defined as follows:

It is readily apparent from this definition that a crime hotspot is a small geographical area (“SGA”) within the larger geographical area of a Basic Command Unit (“BCU”) or an Operational Command Unit (“OCU”). An SGA will be a crime hotspot if the crime rate in that SGA, or the rate at which crime in that SGA is rising, falls within the upper quartile (i.e. top 25 per cent) of the SGAs in the geographical area of the BCU or OCU in which the SGA is located.

Ms Chiswick’s second witness statement makes clear that the SGAs are hexagons superimposed on the map of London to form a grid. She describes the grid as follows:

In order to determine whether any particular hexagon should be treated as a crime hotspot, each hexagon is scored according to three variables. As Ms Chiswick’s second witness statement explains:

Returning to the terms of the Policy, as we have mentioned, the second category of Use Case A is a “missing person hotspot”, which is a SGA in which MPS intelligence or operational experience indicates missing persons are likely to be present (para.2.3(b)). The formation and extent of missing person hotspots was not the subject of challenge by the claimants.

It is an important constraint that each Use Case has particular and specified “relevant policing objectives.” In relation to Use Case A, the relevant policing objective is “to support hotspot policing” but “only for the purposes of locating those Sought Persons who have been placed on the relevant LFR Watchlist in accordance with the watchlisting principles… in section 4 of this policy” (para.2.5, emphasis in the original).

A Central Tasking and Co-ordination Group chaired by a Deputy Assistant Commissioner will consider bids for policing resources. The Group may allocate LFR resource to a hotspot location subject to (1) completion of the relevant form, called an LFR Form 1, and (2) the other requirements of the Policy including (as specifically stated) the requirements as to permitted locations for LFR (para.2.4).

Paragraph 2.6 of the Policy states:

Relevant Hotspot Offence Types are defined in Annexes A and G.

Under Use Case B, LFR may be deployed to support “protective security operations” (“PSOs”). Paragraph 2.7 provides:

By way of example, LFR was used at the Notting Hill Carnival in August 2025 (an “event PSO”) and for a number of hours at Heathrow Airport in November 2025 (a “CNI PSO”).

The relevant policing objective of Use Case B is to locate persons on a PSO watchlist. LFR may also be used to deter or disrupt the attendance of those who pose a threat to CNI or public safety, as the case may be (para.2.8).

Under Use Case C, LFR may be used to locate people based on specific intelligence. As set out in para.2.9:

In the context of this Use Case, LFR can only be used for the purpose of locating the “relevant watchlisted person” (para.2.10).

As we have already indicated, it is an important further constraint that, unless the deployment of LFR falls within one of these three Use Cases, LFR technology is not permitted (para.2.2).

The inclusion of Sought Persons on any particular watchlist is “responsive to the particular use case being considered for LFR Deployment” (para.4.1). It follows that there must be a connection between the persons on any particular watchlist and the use to which the watchlist will be put.

The persons liable to be included on a crime hotspot watchlist are set out in paragraph 4.3 of the Policy as follows:

In so far as the persons who may be included on a crime hotspot watchlist are constrained by reference to those associated with “Relevant Hotspot Offence Types”, this term is defined in Annex A of the Policy as meaning “relevant offences that have been committed or are likely to be committed in the crime hotspot by reference to crime data and/or MPS intelligence reporting and/or operational experience.” Annex G lists the relevant offence types namely: theft and kindred offences; sexual offences; violent and/or weapons related offences; drugs offences; terrorism offences; bail offences; fraud and dishonesty offences; damage to property offences; firearms offences; immigration offences; and modern slavery, kidnap, human trafficking and kindred offences.

As regards a missing person hotspot, the Policy states that “All missing persons sought by the MPS will be added to the LFR Watchlist for a Missing Person Hotspot” (para.4.4).

The categories of persons who may be placed on a PSO watchlist are specified in detail at paragraph 4.5:

There was and could be no challenge to the accessibility or foreseeability of this part of the Policy.

As regards specific intelligence, paragraph 4.6 of the Policy lists those categories of persons who will be added to an LFR watchlist in respect of any Use Case where the MPS has specific intelligence which indicates that a person falling within one of the categories is likely to be in the area of the proposed LFR deployment. The categories are again detailed and specific, namely:

Once again, there was and could be no challenge to the accessibility or foreseeability of this part of the Policy.

Section 5 of the Policy deals with the locations where the use of LFR is permitted. It states as follows:

Section 6 of the Policy is entitled “Proportionality” and deals with Convention rights. The section acknowledges that LFR affects the Convention rights of members of the public, including rights protected by Articles 8, 10, 11 and 14 of the Convention. Section 6 provides practical guidance to police officers on the application of Convention rights and proportionality principles to the types of circumstance that they are likely to encounter in the deployment of LFR. Mr Squires confirmed that no criticism is made about the content of section 6.

It is mandatory for those who authorise LFR deployments to consider whether any particular deployment would be a “proportionate means of achieving the MPS’ policing objectives” (para.6.1). If the proposed deployment is not considered proportionate, it is mandatory for authorisation to be refused (para.6.2).

Any interference with Convention rights is the “starting point” for the consideration of whether LFR may be deployed (para.6.4):

Paragraph 6.5 continues:

The Policy requires decision-makers to consider proportionality in three stages which we summarise. Under Stage 1, the Authorising Officer must be satisfied that the deployment would be likely to achieve one or more relevant policing objectives, based on crime data and/or MPS intelligence and/or operational experience showing why the proposed deployment falls within one of the Use Cases (para 6.7). Stage 2 requires consideration of whether there are alternative means of achieving the relevant policing objective that would have a lesser adverse impact on Convention rights (para 6.8). If the relevant policing objective could be achieved by less intrusive methods, the deployment “should not be authorised as it will not constitute a proportionate means of achieving the MPS’ legitimate aims” (para 6.9).

Stage 3 requires the overall consideration of whether the deployment strikes a “fair balance between any interference with the rights and freedoms of those affected by the Deployment and the achievement of relevant policing objectives” (para 6.10). The balancing exercise “requires consideration of all the circumstances” including (1) the degree to which the deployment will enable policing objectives to be achieved; (2) the degree to which the deployment supports fundamental rights by providing reassurance to those exercising their rights that they can do so safely; (3) the effectiveness of LFR as opposed to other alternatives; and (4) the degree to which LFR would interfere with fundamental rights (para 6.10). The Policy makes clear that, in any particular case, the more serious the interference and the greater the number of people whose rights are affected, the less likely that LFR would be proportionate (para 6.10(d)).

Section 7 of the Policy deals with the specific roles and duties of the police officers who carry out LFR deployments: LFR Operators, LFR System Engineers, LFR Engagement Officers and LFR command roles (paras 7.1-7.4). The Policy lays down requirements for the training of those Operators, Engineers and Officers and their responsibilities in relation to deployment of LFR Para. 7.2)

A Gold Commander has strategic command of the particular LFR deployment and must ensure both that the deployment is carried out in accordance with the objectives given in the LFR Form 1 and that the use of LFR remains necessary and proportionate (para.7.4(a) – (b)). A Silver Commander has frontline tactical command and must ensure that the deployment stays within the assessments that supported it (para.7.4(d)). One or more Bronze Commanders are assigned operational command responsibilities (para.7.4(e)). A Bronze BCU Supervisor will oversee and manage community relations having regard (among other things) to the Community Impact Assessment (para.7.4(e)). This clear chain of command is a far cry from the whim, caprice, malice or predilection of an individual officer.

Section 7 of the Policy makes plain that the generation of an alert should not lead automatically to someone being stopped by the police or to a person’s arrest:

Section 8 deals with standards of governance and oversight in relation to the use of LFR technology. We note that it contains significant safeguards against the arbitrary deployment of LFR. In order to seek authorisation for LFR, an officer must apply in writing to the Authorising Officer using the LFR Form 1 (para.8.2(a)). Specified assessments must be provided including a Community Impact Assessment, an Equality Assessment and a Data Protection Impact Assessment (para.8.2(g)).

During the deployment, logs must be completed to record the “planning and execution of the Deployment” (para.8.3(a)). The Silver Commander must ensure that the deployment remains “necessary and proportionate for the policing purpose identified” (para.8.3(b)) with “absolute discretion” vested in him or her “to suspend or terminate the Deployment” (para.8.3(c)).

Internal oversight of LFR as a whole is provided by a Technology Board which is answerable to the MPS Management Board (para.8.6(a)).

Section 9 deals with data processing and data security issues. It was not suggested by the claimants or the Commission that the terms of the Policy do not meet the requirements of data protection laws.

Section 10 deals with publicity and information about LFR use. It suffices for present purposes to quote the following passage, which deals with how members of the public – including the claimants – may learn about where LFR will be or is being deployed:

A series of annexes to the Policy provides further guidance on various topics, including the use of LFR where it affects children, the disabled and those undertaking gender reassignment where particular privacy considerations are raised (Annex D).

There are other documents that complement the Policy. We were provided with a Data Protection Impact Assessment dated 11 September 2024; a Data Protection Appropriate Policy Document; an Overt LFR Legal Mandate which deals with (among other things) Convention rights and the Equality Act 2010; and an Equality Impact Assessment.

We were shown a blank copy of the LFR Form 1 that must be used to obtain authorisation for LFR deployment at a particular location and time. The application form is in two parts. In the first part, the person making the application is guided through various steps required to ensure compliance with the Policy. In the second part, the Authorising Officer (who will consider the content of the form and decide whether to authorise the deployment) is guided through the relevant considerations including interference with Convention rights and proportionality.

The risk of false alerts – matching an innocent person to a Sought Person on a watchlist – is related to the “similarity score” that we have mentioned above. As we understand the evidence provided by the MPS, the score is a number which has always been 0.6 or more as representing the threshold for a technically-assured and high degree of accuracy.

In 2023, the MPS commissioned the National Physical Laboratory (“NPL”) to conduct an equitability study of MPS facial recognition technology. The NPL evaluation took place after the event in a test environment using footage taken from real life deployments. The NPL Report (completed in March 2023) records that, in tests that used a 0.6 similarity score, the false positive identification rate was 0.017% for a watchlist size of 10,000 (1 in 6000) and 0.002% (1 in 60,000) for a watchlist size of 1,000.

We were not referred to the text of the Report but Ms Chiswick’s second witness statement describes it as showing that there were no statistically significant differences in true positive identification rates and false positive identification rates by gender or race. There was no challenge to that description and no evidence to contradict it. Overall, there was no challenge to the NPL methodology or to its conclusions under test conditions.

Turning from test conditions to real life operations, the similarity score has not remained static. From January 2020 until June 2024, the number was 0.6. From 11 July 2024 to 23 July 2024, it was 0.62. Since 25 July 2024, the similarity score has been 0.64. The higher the threshold number is set, the lower the risk of generating a false alert.

According to a helpful Note from Mr Hogarth drafted at our request, summarising the relevant pleadings and evidence, in the period 1 January to 10 September 2024, there was one false alert for every 33,435 faces that came within a Zone of Recognition and 2.1% of total alerts were false alerts. From 11 September 2024 (when the Policy came into effect) to 31 December 2024, there were 453 alerts generated of which none was a false alert. From 1 January 2025 to 10 October 2025, there were 12 false alerts of which four led to no engagement at all between the MPS and the subject of the false alert.

As we have indicated, Mr Thompson is a voluntary worker for an organisation called Street Fathers that aims to prevent children and young people from coming into contact with the criminal justice system. In his witness statement, he explains that, as part of his work, he participates in “patrols” of the streets of London. The patrols aim to disincentivise young people from becoming involved in gang-related or dangerous activities.

It is not in dispute that, on 23 February 2024, Mr Thompson passed an LFR deployment in the area around London Bridge. His face was matched to an image contained on the watchlist in use on that day, so as to generate an alert. The LFR system had detected a high level of facial similarity between images obtained of Mr Thompson and the image of his brother who was on the watchlist as being the subject of an outstanding arrest warrant issued by a court for breach of bail following an allegation that he had caused grievous bodily harm to Mr Thompson. The high level of similarity generated the alert.

Police officers spoke to Mr Thompson. They ceased their engagement with him after considering information held on the Police National Computer about his brother. Mr Thompson was able to show police a copy of his passport photograph on his phone, which enabled his identity to be confirmed.

In his witness statement, Mr Thompson says that, as a result of being mistakenly identified, he regards LFR as depending on “flawed technology.” He is concerned that “the police will be able to scaremonger” innocent people “into giving over their personal data, which will then be held on a police database with no justification.”

Mr Thompson is concerned that LFR cameras are often deployed in places with larger Black communities which will result in disproportionate and intrusive targeting of Black people. He is scared of being wrongly identified by LFR again.

We should record, however, that the claimants do not raise any issues in relation to the Public Sector Equality Duty under s.149 of the Equality Act 2010.

As we have said Ms Carlo is the Director of Big Brother Watch, an NGO that campaigns on privacy issues. Prior to joining Big Brother Watch in 2018, she led policy work on surveillance and technology at Liberty, the human rights NGO. Among other roles, she serves on the Advisory Board of the World Ethical Data Forum and is a member of the Government’s One Login Inclusion and Privacy Advisory Group.

Ms Carlo lives in South London and works near Westminster, travelling round London for work engagements. She says that LFR has been used in many of the areas that she visits for work, social events and shopping, such as Westminster, Oxford Circus and Clapham Junction.

Ms Carlo has provided five witness statements over the course of the proceedings. In her first witness statement, she says that she is “recognised as an expert on the impact of LFR on rights and liberties.” We have no reason to doubt that she has accumulated a great deal of knowledge about the practical operation of LFR in London. However, Mr Squires did not advance her evidence as expert opinion evidence and her statements do not demonstrate that she has assumed any of the duties owed by an expert witness to the court. For example, it is a general requirement of expert evidence that experts should assist the court by providing objective, unbiased opinions on matters within their expertise, and should not assume the role of an advocate (CPR PD 35 para.2,2) By contrast, Ms Carlo plainly adopts the role of an advocate against the Policy under challenge.

As Ms Carlo cannot be treated as an expert witness, there is no basis for the court to consider her evidence on matters of opinion as opposed to matters of fact. Numerous passages of her first witness statement – indeed the majority of the statement – contain her opinions on LFR. Those passages are inadmissible and cannot advance her claim.

Turning to the factual evidence contained in the first statement, Ms Carlo says that she has monitored “countless” deployments of LFR in London since 2017. She witnessed Mr Thompson’s misidentification at London Bridge. She expresses concerns about the “chilling effect” of LFR deployment at political protests and demonstrations. She does not wish to attend protests or other events where LFR is used. She was active in a republican campaign group which organised a demonstration at the King’s Coronation but decided not to attend the demonstration following the announcement by the MPS that LFR would be used at the Coronation. She avoids the Notting Hill Carnival for fear of being captured on LFR cameras.

Ms Carlo’s second witness statement concerns interim applications considered by the court at an earlier stage of proceedings. We do not need to deal with it.

Ms Carlo requires, and has applied for, the court’s permission to rely on her third witness statement. We have considered it de bene esse for the purpose of deciding whether to admit it.

The vast majority of the third statement is commentary and the expression of opinion. Ms Carlo mentions the connection between crime hotspots and areas of heavy footfall, elements of Professor Utley’s second report, the Government’s definition of CNI, and what she calls “replication of LFR policies across police forces.” Her comments and opinions on these topics range far beyond the proper content of a witness statement.

The sections containing comment or argument permeate the third witness statement. It would be an arid exercise for the court to pick out the several sentences that we would consider to be admissible evidence. Owing to the scale of inadmissible evidence contained within it, we refuse to admit the third witness statement and so we refuse the claimants’ application to rely on it. In any event, the content of the third witness statement does not assist the court in determining the specific legal questions that are the subject of the claim. Even if admissible, we would give the statement no weight.

Ms Carlo’s fourth witness statement relates only to interim matters so that we can turn to her fifth witness statement. Once again, she needs and has applied for permission to rely on her fifth witness statement.

In her fifth statement, Ms Carlo expresses concern that ordinary CCTV cameras in London (which are not presently used for LFR) could in the future be replaced and upgraded to permit LFR deployment. As we have said at [68] above, the court will not speculate about the future. The concern about future use is not relevant to whether the present Policy is in accordance with the law.

Ms Carlo goes on to criticise the way in which the Policy is put into practice. She goes into detail about what she regards as inadequate signage to warn people that they are about to enter a Zone of Recognition and says that social media notifications about forthcoming LFR deployments do not adequately delineate where the LFR cameras will be positioned. These complaints about the practical application of the Policy do not assist the court with the IAWL question that is the subject of the claim.

Ms Carlo goes on to deal with political support for, and opposition to, the use of LFR and the comparative positions in Sweden, Germany and Scotland. None of this is admissible evidence. It is irrelevant to the IAWL question.

In a section of her statement on the use of LFR at protests, she says as follows:

This passage too relates to the operation of the Policy in practice, which is not the subject of the pleaded grounds of challenge for which permission has been granted. Ms Carlo ends by saying:

Overall, Ms Carlo’s fifth witness statement has no real evidential value and adds nothing to our consideration of the specific legal questions that arise in the claim. It is inadmissible. We refuse the claimants’ application to admit it.

The court’s function in judicial review proceedings is to determine the questions of public law raised in the pleaded claim. Evidence will not be relevant in judicial review if it does not engage the issues of public law that are contained in a party’s pleaded case. In judicial review, as in other proceedings, a party’s lawyers have a duty to the court to consider with care and objectivity whether the evidence put before the court – from a client or otherwise – is relevant and, if not, then to advise that it is inadmissible. A client should then heed that advice.

If a party relies on irrelevant evidence, opposing parties are put to needless time and expense in checking the extent to which it falls within the case pleaded. Even where that material appears to be irrelevant, a party may incur further costs considering whether to respond and, if so, how. Some may feel under tactical pressure to do so to avoid a forensic point being made that the material has gone unchallenged. If an opposing party does file evidence, that may then provoke a reply and so the unnecessary proliferation of such material continues. This bad practice wastes the court’s finite resources, for example, in pre-reading and in time taken up at a hearing.

The Commission relies on the witness statement of Ms Angharad Davies, a Principal in its Compliance Team. Since October 2023, she has led the EHRC work on police use of facial recognition technology, including LFR.

Ms Davies explains that the equality and human rights implications of digital services and artificial intelligence have been a key priority for the Commission in recent years. She describes the numerous ways in which the Commission has contributed to public debate and policy-making on facial recognition technology in general and LFR in particular. We do not find her language sufficiently specific to gain a clear picture of what the EHRC has undertaken and achieved. We note however that it has worked with the Home Office, police forces, the College of Policing, the Local Government Association and the Information Commissioner’s Office in relation to the equality and human rights impacts of artificial intelligence. In September 2022, it published guidance called “Artificial Intelligence: meeting the Public Sector Equality Duty”

Ms Davies confirms the Commission’s view that the responsible use of LFR by the police “can play a role in preventing and tackling crime, thereby contributing to public safety.” She observes that “it is essential that clear rules are in place to ensure that [LFR] is only used when necessary, proportionate and with appropriate safeguards.” She emphasises that the use of LFR at protests risks a “chilling effect” on individuals’ rights under articles 10 and 11 of the Convention, which are “central to democratic functioning.” Given the importance of article 10 and 11 rights, the Commission’s view is that “a higher justification for police to use [LFR] at protests is required.” We pause here to note that the EHRC’s concerns may be important but they do not concern the question with which the court is concerned in this claim.

Ms Davies describes how the Commission employs a statistician, Dr Arturo Lonighi, who has carried out statistical analysis of the MPS’s data relating to LFR deployment between 11 September 2024 (when the Policy came into effect) to 10 May 2025. His analysis shows that, for the 130 deployments of LFR in this period, the total number of people whose biometric data was extracted and analysed was 1,707,194. The smallest watchlist size was 13,419; the largest was 19,199; the average was 15,537. According to Dr Lonighi, the average rate of positive identifications relative to watchlist size was 0.06%. The highest rate was 0.16% and the lowest was 0%.

As we have set out above, the claimants seek to rely on the second and third reports of Professor Martin Utley. He is currently a Professor of Operational Research (“OR”) at the UCL Clinical Operational Research Unit having previously served as the Director of this Unit for over ten years. His expertise lies in OR in the health and healthcare fields. He does not purport to have any previous experience in policing methods or crime-related statistics. He says in his second report that the methods used in his research on health apply equally to the issues he has been instructed to consider in the present claim. When we expressed some doubt that a health expert can assist the court on policing methods, Mr Squires clarified that the claimants rely only on his mathematical modelling and not on any wider empirical analysis of LFR usage.

In his second report, Professor Utley used publicly available data to estimate the proportion of each BCU that could be labelled as a crime hotspot under the Policy as at January 2024, and the additional proportion of each BCU where LFR could be deployed at access routes to hotspot areas. He calculated these proportions using SGAs in the shape of squares (not hexagons) for the whole Metropolitan Police District. He concluded:

As we understand it, Professor Utley is here saying that the MPS could, under Use Case A, decide to locate LFR cameras in 85% of the area that makes up the Metropolitan Police District. Relying on this analysis, the claimants’ amended grounds of challenge contend that “it is likely the defendant can choose anywhere he wishes within over 80% of London to locate an LFR deployment as it will be a ‘crime hotspot,’ or contain ‘access routes’ to a hotspot.”

In her second witness statement, Ms Chiswick criticises the accuracy of Professor Utley’s conclusion, pointing out that the physical and practical realities of London mean that LFR technology cannot physically be deployed in such a large part of London. She points out that Professor Utley had made no allowance for private property, such as housing, back gardens and business premises, where the police would have no legal authority to place cameras. She points out that Professor Utley overlooked physical barriers to the positioning of cameras such as the River Thames. He had measured access routes by drawing an additional 300m wide zone around the perimeters of SGAs whereas the Policy is limited to actual access routes, not with all points within a 300m radius of a crime hotspot.

Ms Chiswick describes how the MPS tested Professor Utley’s conclusions by undertaking their own modelling whose conclusions she describes as follows:

Professor Utley undertook further calculations in light of Ms Chiswick’s criticisms. Having been informed that the MPS had reached its conclusion using a map-based filter that enabled modelling in relation to public spaces, he decided to commission analysis from Centre Maps, a company offering mapping and related data analysis. He asked Centre Maps to measure the area within the Metropolitan Police District that is categorised as “roads”, “tracks”, “paths” or “roadside” (together “roads”) using Ordnance Survey mapping. By limiting his analysis to roads, he concluded in this third report that 52% of London lies within a crime hotspot with a further 10% associated with areas deemed “access routes” by the MPS, which gives a total of 62%.

Professor Utley’s figures of 85% and 62% are unenlightening in the context of the issues raised by the present claim. As set out in their Note, the claimants rely on Professor Utley to found the submission that the MPS is able to select virtually any area of high footfall across London and deploy LFR there in the hope that a Sought Person will be present. We do not agree.

First, as the claimants’ Note on Expert Evidence concedes, Professor Utley reaches his conclusions by an examination of only a single, aspect of the Policy: the “where” question as it relates to one part of Use Case A, namely “crime hotspots” (as defined in para 2.3(a)) and access routes to crime hotspots (as defined in para 5.2). He was instructed by the claimants’ solicitors to deal with a second aspect of the Policy, namely “the proportion of London that could be considered to be within the deployable range” of CNI but declined to reach a conclusion on this question for want of information about what constitutes CNI, Accordingly, the scope of his evidence is necessarily limited.

Even within its own narrow compass, Professor Utley’s evidence fails to deal with the significant constraints on the location of LFR for the purpose of crime hotspot policing which are imposed by other sections of the Policy, including the sections on the “why” question and the “who” question. As we have set out above, these constraints include (but are not limited to): (1) the requirement that LFR must only be used to support the relevant policing objectives for Use Case A (para 2.5); (2) the construction of watchlists (which will in turn influence whether to deploy LFR) in accordance with the detailed provisions of the Policy (see especially para 4.3); (3) the mandatory consideration of relevant hotspot offence types as a building block in watchlist construction (para 2.6); and (4) a mandatory proportionality assessment by an Authorising Officer prior to any deployment (para 6.1). Professor Utley’s evidence does not and cannot speak to these constraints which are bound to influence where LFR can and cannot be deployed.

Secondly, even taking Professor Utley’s conclusions at their highest, they mean no more than that LFR cameras may (if other constraints within the Policy are met) be placed in 85% or 62% of a large metropolitan area. This tells the court nothing about whether the Policy is in accordance with the law. A policy may be in accordance with the law if it covers a large proportion of London’s geography. Conversely, it may not be in accordance with the law if it covers a small area. There is no inherent link between the area covered by a policy and its lawfulness. Mr Squires was unable to make any persuasive submissions about how the possibility of using LFR in 85% or 62% of London may demonstrate that the Policy is not in accordance with the law.

Put another way, the geographical proportion of London coverable by LFR is a consequence of the defendant’s policy decision to define crime hotspots by reference to areas where the crime rate and/or rate at which crime is rising is assessed to be in the upper quartile for any particular BCU or OCU area. The defendant’s decision to select the upper quartile means that more of London may be the subject of LFR deployment than, say, a decision to select the upper decile would allow. But the selection of the upper quartile has not been challenged by the claimants. It follows that Professor Utley’s conclusions flow from a policy decision that is not the subject of the present claim. The difference between the use of quartiles or deciles as a measure is not something which conceptually goes to arbitrariness or whim, particularly, when paras. 2.3(a) and 5.2 are read together with the remainder of section 2 and, indeed, the Policy as a whole.

Thirdly, as we have said, Mr Squires relies on Professor Utley’s analysis to impugn the Policy on the basis that it permits the defendant to select virtually any area of high footfall across London. But Professor Utley’s reports do not deal with footfall. The claimant’s solicitors, who commissioned the reports, did not ask him to do so. His instructions did not ask him to consider (for example) whether the terms of the Policy are weighted against scenes of crime in low footfall areas. They did not ask him to provide any analysis of correlation between footfall and crime.

Ms Chiswick accepts in her second witness statement that there is “often likely to be a degree of correlation between footfall and crime”. Areas of high footfall contain “plenty of potential victims, and an environment in which victims may be easily distracted”. She points out too that it is generally easier for criminals to evade detection in crowded areas, and that areas that contain more people will (other things being equal) contain more offenders.

However, Ms Chiswick goes on to make the following powerful observations:

Professor Utley was not instructed to deal with Ms Chiswick’s evidence on footfall, so that we do not know what, if any, response he would have been able to provide. A key part of the claimant’s reliance on Professor Utley therefore revolves around a subject – namely the relation between footfall and crime hotspots – that he was not asked to model. That approach is self-evidently inadequate, shoehorning Professor Utley’s evidence into legal submissions that his evidence was inapt to advance.

Fourthly, Mr Squires agreed that the MPS may be justified in using LFR to apprehend offenders in certain crowded places such as football stadiums. He did not criticise its use at event PSOs “which are expected to attract public attendance” (para.2.7(b) of the Policy). He did not explain why interference with privacy or other rights at football matches and public events with high footfall is capable of meeting the IAWL requirement while leaving too much discretion to officers in other contexts. The claimants articulated no proper grounds for maintaining that there is a connection between high footfall and their allegation that the Policy is not IAWL.

It is the defendant’s function to formulate policy on LFR deployment. It is the function of the court, in the exercise of its supervisory jurisdiction, to ensure that the Policy is in accordance with the law. The court will not rewrite the Policy or compel the MPS to use LFR in any particular way across any particular area. Judicial review is not designed for such a purpose. To the extent that Mr Squires relied on Professor Utley’s evidence to impugn the geographical reach of LFR and its use in crowded areas of London, his submissions missed any proper target for judicial review.

By virtue of CPR 35.1, the court will restrict expert evidence to “that which is reasonably required to resolve the proceedings.” In applications for judicial review, the court has no power to resolve anything other than points of public law. It follows that expert evidence in judicial review proceedings must be reasonably required to resolve a point of public law. Professor Utley’s reports do not assist the court in that endeavour. The application to rely on Professor Utley’s reports is refused.

We reiterate this court’s observations in R (AB) v Chief Constable of Hampshire [2019] EWHC 3461 (Admin) (Dame Victoria Sharp P and Lewis J as he then was) that:

Ms Chiswick’s third witness statement responds to Professor Utley’s third report as well as to several points raised by Ms Carlo’s fifth witness statement. As we have said, Ms Chiswick needs the court’s permission to rely on her third statement which we have considered de bene esse.

We sympathise with Ms Chiswick for feeling the need to respond to Professor Utley and Ms Carlo. However, the MPS position has always been that expert evidence is irrelevant and we have determined that Ms Carlo’s fifth statement is irrelevant too. The response to irrelevant evidence does not necessarily lie in filing further irrelevant evidence. As Farbey J said in argument, Professor Utley and Ms Chiswick have engaged in a game of ping pong which has assumed, in each round of shots, a “momentum” that is less and less related to the issues in the claim. Neither party should have engaged in this fruitless approach. We refuse, therefore, to admit Ms Chiswick’s third statement.

We now turn to the grounds of challenge.

Mr Squires’ overarching contention was that the defendant’s ongoing use of LFR under the Policy is not IAWL (under Article 8) and not PBL (under Articles 10 and 11) because it does not sufficiently constrain where the MPS can, in practice, deploy LFR. As regards the “where” question, the Policy leaves too much discretion to individual police officers which was one of the fundamental deficiencies of the policy considered by the Court of Appeal in Bridges (see paras 91 and 121 of the Court of Appeal’s judgment).

As Mr Squires did not contend that there was any material legal distinction between IAWL and PBL, we are able to deal with grounds 1 and 2 together.

It is telling that, unlike in Bridges, Mr Squires suggested no specific safeguards against arbitrariness that are missing from the Policy. Indeed, many of his submissions were advanced at a high level of generality and did not properly distinguish between the application of the Policy (which is not the subject of the claim) and the issue whether it has the quality of law (which is).

Mr Squires’ submissions, supported by those of the Commission, may be discussed under a number of themes.

Mr Squires emphasised that, under para 2.3(a), a crime hotspot is defined not only by reference to crime data and intelligence reporting but also by reference to “operational experience as to future criminality” (para.2.3(a)). He submitted that the term “operational experience” is too opaque and subjective to meet the test of foreseeability. He submitted that the inclusion of operational experience as one of several factors to be considered in designating a SGA as a “crime hotspot” provides too broad a discretion to individual police officers.

We disagree. The Policy does not refer to “operational experience” in general. Instead it requires that the operational experience regarding future criminality indicates an area where the crime rate and /or the rate at which it is rising is assessed to be in the upper quartile for the relevant BCU/OCU. The criterion is specific. The same is true where “operational experience” is referred to elsewhere in the Policy, for example paras.2.3(b) and 2.6 under Use Case A (repeated in Annex A),

Ms Proops submitted that the term “operational experience” refers to the MPS’s specialist corporate experience of London policing and so was easily distinguishable from the whim or caprice of individual officers that the IAWL requirement was designed to guard against. She submitted that the public would expect the police to apply its specialist operational knowledge and judgment in selecting a crime hotspot for an LFR deployment. The application of corporate experience to decision-making guards against rigid decision-making and the fettering of discretion. The application of specialist judgment ensures that decision-making is not capricious.

We agree with Ms Proops. In their proper context, the words “operational experience” are not vague. They reflect evidence-based judgments made about individual SGAs. They capture the specialist and corporate knowledge of the defendant and his officers who have functional and constitutional responsibility for policing London. They are readily comprehensible.

To the extent that the claimants demand further detail, we have been directed to no authority for the proposition that, in the context of policing, the claimants are entitled to be given a breakdown of the sort of operational judgments that may cause the MPS to select one location rather than another. It would be unworkable (and unrealistic) for the Policy to “provide for every eventuality” by spelling out the multifactorial ways in which operational experience may be brought to bear (Gillan at [77] quoted at [55] above).

It is not realistic to maintain that the MPS should disregard its operational experience when making decisions about where its resources are to be used, and where not. As Ms Proops submitted, in the context of law and order, the public can expect the defendant to ensure that MPS operational experience is not ignored.

This sort of experience is a far cry from the “hunch” or “professional intuition” of an individual officer which concerned the ECtHR in Gillan, or the “whim” and “caprice” which concerned Lord Bingham. The requirement of foreseeability is met.

Mr Squires sought to impugn the legality of the Policy on the grounds that the MPS has used the same watchlist on more than on occasion. Referring to the list of watchlists on the Schedule of MPS Deployments from April 2023 to 31 December 2024, he asked us to infer that the same watchlists are deployed in different locations. This was (he submitted) powerful evidence that LFR cameras were deployed in arbitrary locations in the hope that someone on a watchlist would come past.

The Schedule shows that, in 2023 and 2024, the MPS used watchlists that each contained between around 10,000 and 17,000 names. We agree that these very large numbers mean that the inclusion of two identically-sized watchlists is not easily explicable as coincidence and that, if two watchlists are precisely the same size, the same watchlist may have been used twice for different locations.

However, we are not in a position to reach any conclusion of public law about repeat use because the claim has not been pleaded or prepared in such a way that that we could reliably do so. The repeat use of some watchlists, which the Schedule appears to show, does not advance the claimants’ legal analysis of IAWL or PBL. In so far as the point has any force, it would be relevant to the application of the Policy and not whether the Policy has the quality of law.

It is not in dispute that the court should adopt the “relativist approach”, accepted by the Court of Appeal in Bridges, that the more intrusive the act complained of, the more precise and specific must be the law said to justify it. The Court of Appeal in Bridges at [77] contrasted information obtained from LFR with information obtained by intrusive techniques such as bugging or DNA sampling.

In their written submissions on behalf of the Commission, Ms Dee Masters and Mr Joshua Jackson urged us to undertake a reappraisal of intrusiveness in light of technological developments since Bridges was decided. They submitted that LFR enables “mass surveillance” and the processing of “sensitive” personal data within the meaning of section 35(8) of the Data Protection Act 2018. Personal data of many thousands of people is processed in a penetrating way by the creation and use of biometric templates which are extracted and rapidly compared to a watchlist. The consequences of errors are significant, as Mr Thompson’s experience shows. Ms Masters and Mr Jackson submitted that the context was often public protest and the expression of political opinion which should be protected from any chilling effects.

On the evidence before us – which must underlie our conclusions about intrusiveness – we are not persuaded that the technology or any other material factor has changed significantly since Bridges. As the Divisional Court in Bridges observed at [75], LFR simply involves taking a photograph of a person’s face (along with others), conversion to biometric data and the use of algorithms to attempt to match it with the data for the faces of persons on a watchlist. No physical entry, contact or force is needed when using LFR to obtain biometric data. The Court of Appeal stated that LFR information is not obtained by any intrusive technique such as bugging or DNA sampling whilst recognising that the use of the biometric data is more intrusive than simply capturing images on CCTV ([77] and [85]).

The Commission emphasises that the world has seen the rapid development of artificial intelligence very recently with risks that, unless constrained by domestic and international law, States may use digital technologies to suppress dissent. In the domestic context, the uses to which LFR may be put, and the technology which supports it, will doubtless evolve but this claim is concerned with the terms of the Policy under challenge and (as we have said) its lawfulness now. We see no reason to take a different approach to Bridges.

Any intrusion to which the claimants are exposed by the deployment of LFR is not directed at them in the sense that, save for unintended errors, the MPS has no interest in their biometrics or in engaging with them. We regard this as a further reason why the capture and processing of personal data by LFR software is not at the same end of the “spectrum” of intrusive executive action (to reflect the language of the Court of Appeal in Bridges) as in Malone and other cases cited to us where executive action was specifically aimed at circumscribing or otherwise regulating a particular individual’s conduct.

Mr Squires submitted that the arbitrary nature of where LFR may be deployed under the Policy may lead to disproportionate deployment in areas of London which are lived in by ethnic minority communities.

We heard no developed or meaningful challenge on discrimination grounds. We accept that, if the Policy had the effect of discriminating against a section of the community, it could or would give rise to arbitrary intrusion on individual rights and would not have the quality of law. However, the risk and potential scope for discrimination on grounds of race was no more than faintly asserted by Mr Squires. We are not able to accept, on the thin submissions advanced before us, that concerns about discrimination infect the legality of the Policy.

Mr Squires submitted that the vast number of people exposed to having their images processed by LFR was relevant and the largescale effect on innocent people was critical. The Policy failed to constrain the undue use of LFR for the mass of innocent people. The broad discretion permitted as to where LFR software may be used has a chilling effect on the right to political protest and so should be constrained by the court.

As we have already set out, laws and policies must be sufficiently foreseeable in their terms to give individuals an adequate indication as to the circumstances in which and the conditions on which the authorities are entitled to resort to measures affecting their rights under the ECHR (see Martinez, [117] quoted at [50] above). This does not mean that a public authority must spell out in its published policies every instance and every circumstance in which it will exercise a discretion or power. None of the authorities to which we were directed go so far.

We reiterate that, in this area of the law as in others, context is important. In the words of the ECtHR with which we agree, an assessment of whether a measure has the quality of law involves the court having “regard to the specific purposes of the contested power” (Versaci v Italy, Application No. 3795/22, 15 May 2025, para.119). In certain areas of public policy, absolute certainty cannot be expected or is unattainable. As the ECtHR held in Magyar Kétfarkú Kutya Párt v Hungary (2020) 49 BHRC 41 at [94]:

As held in Malone at [67] (which we have quoted in [47] above), the concept of law must be sufficiently clear in its terms to give citizens an adequate indication as to the circumstances in which and the conditions on which public authorities are empowered to interfere with the rights of citizens. In the context of promoting law and order in a large metropolis, the Policy provides the claimants with an adequate indication of the circumstances in which LFR will be used and enables them to foresee, to a degree that is reasonable in the circumstances, the consequences of travelling in an area of London where LFR is in use.

Mr Squires’ main submission appeared to reduce to the proposition that, in relation to LFR deployment in crime hotspots under Use Case A, there is no restriction on the MPS using LFR even where there is no connection between the persons being sought (ie the “who” question) and the location of the deployment (ie the “where” question). He submitted that there was nothing to prevent a person who lived and committed crimes in South West London from being placed on a watchlist for an LFR deployment in North East London. He submitted that the principal constraint relied on by the defendant, namely the proportionality requirements in section 6 of the Policy, is no constraint at all because it amounts to no more than a statement of the law with a request by the relevant officers to follow it. Mr Squires invited us to conclude that there is little or no more meaningful constraint on the MPS selecting any crowded place they wish than there was in Bridges.

Ms Proops submitted that Mr Squires had taken section 5, which deals with the “where” question, out of context. She submitted that, when read as a whole, the various interlocking sections of the detailed Policy contained clear, precise and mandatory constraints on where LFR may be deployed. She submitted that the “who” question and the “where” question were rationally linked. The overarching requirement of proportionality, as set out in detail in section 6, was a meaningful and effective safeguard against the arbitrary deployment of LFR.

We agree with Ms Proops that the Policy contains adequate and lawful constraints on where LFR may be deployed. First, the “where” question is linked to the “who” question to a significantly greater extent than Mr Squires suggested. Secondly, as Ms Proops submitted, the “why”, “who” and “where” questions are related and form interlocking parts of a Policy that does not authorise the making by the police of arbitrary decisions, but rather decisions based on a policy which is IAWL and PBL. Thirdly, an overarching constraint on all elements of the deployment of LFR is provided by the mandatory requirement of proportionality as set out in clear terms in section 6 of the Policy.

These three aspects of the Policy defeat Mr Squires submissions, for the following reasons.

A person will be added to a watchlist on suspicion of criminality that amounts to one or more Relevant Hotspot Offence Types for the crime hotspot in which LFR is to be deployed (para.4.3(a)). As Ms Proops submitted, there is thus a link between who is on a watchlist and the crime hotspot (ie the area) in which LFR will be deployed. There is a further link between “who” and “where” in that a person will be added to a watchlist if he or she has committed (or is committing or is about to commit) an offence that is a Relevant Hotspot Offence Type for the area of the planned LFR deployment.

A similar analysis applies, for example, to those who are placed on watchlists because they are subject to court orders (para.4.3(d)). For this category of person, it is a requirement for entry on a watchlist that the offence or offences for which they have been charged or convicted are of a type amounting to one or more Relevant Hotspot Offence Types for the crime hotspot in which the LFR deployment will take place (with similar provisions for those subject to civil orders not made in criminal proceedings).

Ms Proops accepted that the other categories listed in the “who” section of the Policy are not linked to the location of an LFR deployment: serious criminals (para.4.3(b), those wanted by the courts (para.4.3(c), and MAPPA offenders (para.4.3(e)). For these categories, however, there is overall protection against arbitrariness by virtue of other requirements of the Policy, namely the “why” question and the proportionality requirement.

We have set out above the important aspects of the “why” question. Not least, as set out in paragraph 2.1 of the Policy, the purpose of LFR is to locate people. By virtue of paragraph 2.5, LFR may only be used in the context of crime hotspots for the purposes of locating Sought Persons on a watchlist. This means that an application for resources for the deployment of LFR that is not designed to locate people will fail.

But it also means that, provided that the watchlist is capable of locating some people, it may contain the names of other people (such as those wanted by the courts where the Policy does not require a connection with the location of LFR) with no more or less intrusion on the human rights of innocent people than if those other names were not placed on the watchlist. Whether this is an effective style of policing is another question which we have not been asked to consider and is not in itself a question for this court. There is however no lack of foreseeability.

Furthermore, proportionality is an overarching constraint on all aspects of LFR deployment. As we have said, Mr Squires submitted that the section of the Policy on proportionality is no real constraint as it amounts to no more than a statement of the law. We reject that submission. As we have said, section 6 gives guidance that is specifically and practically linked to the sort of situations that face the police when determining whether to deploy LFR. It also provides a clear and structured approach for decision-makers.

As regards the potential interference with privacy rights, Article 8 is considered in detail with practical examples. The Policy recognises that Article 8 “may be engaged with greater intensity if the proposed Deployment is to an area in which members of the public have greater expectations of privacy, for example close to a clinic or a school”. A sporting facility “may attract a greater expectation of privacy when it is being used as a private members’ club than if it is used to host a major ticketed sporting event” (para.6.4(a)).

In relation to interference with Article 10 and 11 rights, the Policy expressly requires the police to “have regard to the risk that deployment will have a chilling effect on the willingness of some individuals to take part in lawful assembly and express their views” (para.6.4(d)). There is clear and detailed guidance at paragraph 6.5 that enables decision-makers to understand whether any interference with rights is warranted in the particular circumstances with which they may be confronted (see [121] above). There is a structured, three-stage process for the assessment of proportionality (see [122]-[123] above). Of particular importance in the context of the issues raised by the present claim, the Policy recognises the concerns about sheer numbers by stipulating that, in any particular case, the more serious the interference and the greater the number of people whose rights are affected, the less likely that LFR would be proportionate (para.6.10(d)).

In our judgment, read fairly and as a whole, the guidance to officers on proportionality “addresses the factors to be considered” (Christian Institute at [97]). It recognises and deals with the risk of a chilling effect on aspects of public life. It acts as an effective safeguard against arbitrary outcomes.

This is a long way from Bridges in which the Court of Appeal criticised the use of LFR in South Wales when it was not even clear who could be placed on the watchlist or that there were any criteria for determining where LFR could be deployed. It was in these circumstances that the Court concluded that “too much discretion” was left to individual police officers. Neither those pitfalls, nor anything like them, exist in the Policy before us.

For these reasons, the Policy is IAWL and PBL. The claimants’ human rights have not been breached whether under article 8 (ground 1) or under articles 10 and 11 (ground 2). Both grounds of challenge fail.

Ms Proops submitted in the alternative that the court should refuse to grant relief, on the grounds that any defect relating to the quality of law was immaterial and academic. It was highly likely that the outcome for the claimants would not have been substantially different if the conduct complained of had not occurred (section 31(2A). Given our other conclusions, we do not need to consider this aspect of Ms Proops’ submissions.

Accordingly, the claim is dismissed.