Appendix B: National Conversation on Live Facial Recognition – Focus Groups Key Findings

Introduction

The Scottish Police Authority and Police Scotland held five focus groups during April 2025 as part of the National Conversation on Live Facial Recognition (LFR). We also held individual discussions with some organisations who were unable to attend a focus group. In total we engaged with representatives from 27 organisations representing a variety of sectors (e.g. criminal justice, equalities, victims support services, academia etc.). A full list of the organisations who participated is available at the end of this paper.

In total, four main themes became apparent in the discussions. These were centred around:

Assurance

Legislation

Deployment conditions

Support

This paper outlines the subthemes identified underneath the four main themes. Given the cross-cutting nature of the discussions, and the many facets to this topic, there will be some overlap between sections.

Methodology

The notes from the focus groups and discussions were collated and thematically analysed to identify the main overarching themes. These main themes were then thematically analysed to identify subthemes. Both the main themes and subthemes were subjected to intercoder reliability testing.

Results

Assurance

Underneath the main theme of assurance, there were four subthemes identified.

Transparency

Transparency was a key point across all focus groups. It related primarily to how LFR would be operated and how that was communicated accessibly to the public and those subject to LFR (e.g. notification to those in an area subject to LFR, the rationale for those being on a watchlist, who the information garnered via LFR is shared with etc.). It was clear that there needed to be transparency around what police would be allowed to use LFR for and what they were not allowed to use it for to prevent mission creep and to manage expectations of victims. If LFR was used in a way that departed from a legitimate use, then documentation
justifying that departure was noted to be needed. The need for clarity on the operational necessity for LFR as opposed to other policing methods was also expressed. A clear admission and acknowledgement of potential
issues with the technology, along with an assurance that they were being addressed, was also highlighted (e.g. limitations, potential bias etc.).Regarding deployments, there was an interest in having oversight around the cumulative impact of deployments, not just on individual deployments. There was also the suggestion of a central register documenting the use of LFR. A request for Equality and Human Rights Impact Assessments (EqHRIAs) for all deployments was also raised.

Governance and Accountability

Governance and Accountability related largely to guardrails that could be applied should Police Scotland decide to implement LFR at some point in the future. There were discussions around having an independent body for authorising use of LFR (e.g. with experience in human rights) along with the potential for courts/judicial authorisation for LFR deployment, rather than just internally in Police Scotland. If authorisation was internal to Police Scotland, it was stated it would need to be a clear and transparent process. Statutory duties were raised, such as the role of the Public Sector Equality Duty (PSED) requirements (e.g. EqHRIA) along with Data Protection Impact Assessments (DPIA) in the implementation and governance of LFR. It was noted there would be a need to retain the robustness of the authorisation process to prevent it becoming a ‘box ticking’ exercise. There was also caution around created new layers of scrutiny that may not be required or add value depending on the frequency of LFR use.

As well as authorisation, the need for an independent body to review the use of LFR and the appropriateness of deployment (e.g. effectiveness and efficiency) and for independent scrutiny by those with human rights/data protection expertise was also expressed. It was highlighted that post hoc reflections and processes for accountability during deployment, along with a written rationale or log for use, was also important. An annual report by an independent body on usage and review of LFR (with a human rights
focus) was suggested, with mention of corresponding powers should concerns be raised in the report. A right to redress with accessible procedures was also stated to be needed.

There was also discussion around utilising an independent group to consider the adoption of technology like LFR, e.g. the Police Scotland Independent Ethics Advisory Panel, along with developments in this space
outwith Scotland.

Bias Mitigation

The potential bias with LFR technology and its mitigation became a key point in terms of barriers to assurance. It was made clear by participants that there would need to be trialling, and investigation of any
technological solution adopted to provide LFR to ensure it did not contain bias and that the Public Sector Equality Duties were being met. While a ‘human in the loop’ (whereby an officer is involved prior to an approach of someone identified via LFR) was mentioned as a possible safeguard, it was highlighted that human error was also a possibility and it was important to address this.

While there was the report from the National Physical Laboratory, it was highlighted that it was not peer reviewed and other reports note a difference in performance of LFR technology depending on ethnicity. Therefore, it was stressed that there would need to be additional scrutiny and acknowledgement of potential bias and issues, with a point raised that this should be made available to regulatory bodies to ensure statutory duties are being fulfilled.

Purpose

Having a clear purpose for LFR was raised in many of the focus groups. Similar to the subtheme of transparency, this included being clear on the operational need for LFR (e.g. a prescriptive list of offences LFR can be deployed for) and what it was achieving that conventional policing could not (e.g. could the same outcome be achieved via other means). There was caution that if there was a list of offences that LFR could be used for, then historically these lists have expanded over time. It was also made clear that there would need to be evidence to justify the use of LFR, with no usage where the sole justification was that it would be helpful to law enforcement.

Legislation

Under the Legislation theme, five subthemes were identified:

Primary legislation

Democratic basis

Police and Fire Reform (Scotland) Act 2012

Trust/Confidence

Other legislative/regulatory frameworks

Primary legislation

Across several focus groups, the lack of primary legislation and a current lack of legal basis governing the use of LFR was raised as a concern by participants, with the suggestion that no move to introduce LFR is considered prior to the introduction of primary legislation.

The need for primary legislation was viewed as being essential due to the potential for interferences with right to privacy, freedom of expression and assembly, and how this may have a chilling effect with regards to
protests. Primary legislation was viewed as being important due to its novel nature in a policing context, combined with its potential for intrusion, and that a higher threshold must be met legislatively to oversee
its introduction. Some also raised that a lack of primary legislation may lead to an increased likelihood of legal challenge and an erosion of public trust.

The case for primary legislation also referenced the Protection of Freedom Act (2012) that regulates fingerprints, with the view that LFR also warrants similar legislative guidance.

Some also mentioned sunset clauses in legislation, but raised concerns with regards to the efficacy of such clauses. There was also cautioning that primary legislation would require further consideration in relation to
the impact on special category data (e.g. children and young people). Some also suggested that any legislation or code of practice introduced should govern the broader use and exploration of technology in policing. This lack of legal basis was highlighted with regards to how this lack of legal basis may not satisfy an intrusion on human rights, with one participant emphasising that such protections exist to protect minorities.
In particular, some raised concerns as to how other pieces of legislation, such as the Human Rights Act, may be used to justify the use of LFR.

Some suggested that Police Scotland speak to Scottish Government and UK Government to introduce primary legislation.

Participants also raised concerns that if Police Scotland produced its own guidance on LFR use that this would not be sufficient protection for the use of LFR.

It was understood by some participants that there may be a decision by Westminster to introduce legislation that would govern the use of LFR in policing that may make for a clearer landscape. This also related to
ongoing court proceedings in England and Wales, with some participants suggesting that any decision to progress should await the outcome of these proceedings.Some argued that a framework to support LFR could be made through existing legislation or frameworks.

Democratic basis

Participants also raised the lack of parliamentary oversight of LFR and that it is inappropriate to be introduced as there is no democratic mandate and that the use of LFR is a political decision. Some participants also noted the Justice and Home Affairs Committee in the UK Parliament stated that there is an insufficient democratic basis for LFR.

Some participants emphasised the need for a national conversation at Scottish Parliament level given how significant a shift this would be in policing and its potential for intrusion of human rights and private life.

Police and Fire Reform (Scotland) Act 2012

Focus groups considered the provisions within the Police and Fire Reform (Scotland) Act 2012 and the unique legislation in Scotland that considers wellbeing of people of Scotland, which is different to other jurisdictions.
Some argued that Police Scotland has a duty to consider new and emerging technology (even beyond LFR) in pursuit of delivering the best service in line with statutory duties.

Trust/ confidence

The lack of primary legislation was argued by some as having a potentially negative impact on public trust and confidence, particularly where this leaves Police Scotland vulnerable to legal challenge. To positively influence public trust, one participant suggested raising public awareness of how LFR may impact different groups would be positive.

Other legislative/regulatory frameworks

Throughout the five focus groups, discussions covered other legislative or regulatory frameworks that may be supported, or be negatively impacted, through the use of LFR.

This included frameworks included the Scottish Biometrics Commissioner’s Code of Practice, and ensuring compliance with other legislation through completion of relevant Impact Assessments.

Other frameworks such as Regulation of Investigatory Powers (Scotland) Act 2000 (RIPSA) were emphasised as being robust frameworks already in place that Police Scotland must adhere to and that could provide
assurance. In addition, some legislation that Police Scotland were encouraged to consider included the EU AI Act. Some also mentioned the impact on the PSED and that authorities need to give ‘due regard’ but felt
that the implications of LFR would go beyond that requirement. Broader conversation related to pieces of legislation such as UK General Data Protection Regulation (UK GDPR) and ensuring technology passes any strictly necessary threshold. It was noted by some that policing would need to satisfy a purpose test, namely: is there a legitimate, clear policing purpose for using this technology and is it strictly necessary? Participants
noted that any use of LFR would need to consider the impact on these areas to assess whether LFR deployment is appropriate.

Deployment conditions

The following subthemes were identified under deployment conditions:

Oversight

Proportionality

Lawfulness

Transparency

Parameters for use

Oversight

The subtheme under deployment conditions that was discussed in most detail was oversight. Participants referenced the importance of completing key documents, such as an EqHRIA to ensure the impact on individuals is considered. Some also mentioned the importance of gathering appropriate data and insights are utilised where LFR is deployed.

A key aspect of the discussion ensuring the correct processes are in place for the authorisation of LFR – whether this would be through considering new layers of scrutiny or through existing mechanisms. In particular, this related to ensuring there is appropriate seniority of the police officer authorising the use of the technology and when in the investigative process this would be considered. Participants discussed whether there should be an additional level of authorisation or approval that should be implemented, with some of the view that additional approval levels should be implemented, and others cautioning that existing governance could be utilised. It was noted by one participant that this sign off can often not be as robust as it should be, and that this should be considered.

Authorisation also considered who in policing would make the decision to deploy LFR when known risks against specific groups existed. Some participants were of the view that authorisation of LFR use should not be done solely by Police Scotland or overseen by the Scottish Police Authority – rather, this deployment decision would be made through judicial processes or independent experts.

Participants also raised that Police Scotland should make clear how LFR will be approved for use, and whether this will be on a case-by-case basis or through guiding principles.

Human-in-the loop decision making was also viewed as a key aspect of LFR use to reduce an overreliance on the technology. This also included ensuring that Police Scotland does not limit decision makers to police
officers but to also expand this to staff.

Oversight also comprised of discussion on ensuring the performance of the technology and that LFR is adequately tested to ensure its accuracy is sufficient, and how to manage bias and misidentification. This was
considered particularly relevant with regards to reported racial inaccuracies that have been reported for LFR technology and inbuilt biases of LFR. As such, it was viewed as important to provide the public with information on how the decision to use LFR is made. Some called for a more scientific approach to understand how effective LFR is.

The importance of maintaining adequate logs and/or rationale for the deployment of LFR was also mentioned, along with the overall auditability of the processes and ensuring adequate data and insights are captured.
Oversight also considered the importance of defining the use of LFR and how information relating to individuals not on watchlists is managed. The importance of providing adequate oversight was viewed as being vital in maintaining public trust and confidence.

Broader guardrails with regards to oversight considered how use cases may expand and how LFR may be used for broader purposes than originally intended, with some noting that there is a prescribed list of
offences, however historically this can expand. This related to broader points on the clarity of the purposes of LFR, and who will evaluate its performance and deem whether deployments are achieving desired
outcomes. Others detailed the importance of Codes of Practice rather than use cases for LFR.
This also broadly related to wider surveillance processes, such as RIPSA, and ensuring appropriate safeguards are in place. 

Proportionality

Participants discussed the importance of proportionality in the context of deployment conditions. Some explained that they believed that LFR could be used, but that it should be used proportionately. This particularly related to where LFR may be used in the context of ‘serious criminality’, however participants cautioned against scope creep due to the potential impact of this on proportionality of LFR.

Several participants queried the difference in terms of benefits of using LFR in comparison to a standard policing approach and whether LFR would be a proportionate alternative. In addition, this also related to the
number of people arrested through LFR and whether this warranted the level of intrusion that is caused by LFR. The proportionality of LFR was also discussed with regards to its impact on human rights (specifically
freedom of expression, privacy and discrimination) and whether this was sufficient grounds for its use.

Participants specifically referred to legislation such as the Human Rights Act (1998), and the EU AI Act, which also highlights the permissible crimes that would mean the use of LFR was proportionate. One participant also raised the Information Commissioner’s view of police use of LFR, and that it should not be used purely down to its availability, efficiency or capacity to incur savings. One participant raised that lawfulness and proportionality are interrelated in how they impact public trust. The importance of a robust ‘proportionality test’ was emphasised.

Lawfulness

Another subtheme that emerged under deployment conditions was the lawfulness of LFR.
Some participants noted that they understood that LFR could provide benefits as a policing tool, provided its lawful use. Participants cautioned that RIPSA guidelines must be adhered to. Lawfulness also considered the
impact of LFR on vulnerable members of the public and whether this could provide some benefits in some cases, but also that it may have implications in how it impacts groups such as children and young people.

Some also highlighted a report published by the Justice Sub-Committee that there was no justifiable basis for Police Scotland to invest in LFR due to concerns in racial and gender bias, and overall inaccuracy.

Transparency

The transparency subtheme considered the importance of engaging with the public prior to deployments to ensure public awareness and understanding with regards to what LFR is. In particular, there should be
Policing Performance Committee transparency and openness with the public about how LFR is used and the
criteria for its use, why it is being used by policing, and making it obvious to the public that it is in use, with specific attention paid to ensure that this communication is accessible to all. For example, some participants
questioned whether LFR would be in permanent use or whether it would be at specific points. Transparency should also relate back to oversight and assuring the public that there are appropriate safeguards in place.
This also included providing detail to the public on how the technology is used and how the information gathered is used by policing. Transparency also broadly related to accountability and auditability of processes.

Parameters of use

Key to the discussions on LFR was clear parameters for its use. Groups varied on whether they believed an overarching code of practice with principles guiding use or the development of distinct use cases would be
most appropriate, as some stressed that other police use of LFR does not have clear guidelines. However, some raised the point that Police Scotland may not wish to constrain itself by detailing narrow use cases
from the beginning. Some raised that this may result in scope creep, which could impact trust. It was highlighted that particularly with regards to other uses of LFR, the uses should not continue to expand and change, and the thresholds around use.

Parameters of use also considered the guidance on the crime types it is used for, with some suggesting that it is not used solely in instances of major crime, which relate to the overall use cases and code of practice.
However, others raised concerns about LFR being used in broader circumstances, particularly where LFR may be used with regards to children and the potential for overcriminalisation of children. Furthermore,
some suggested that LFR should be considered through the lens of harm to individuals, as opposed to the crime type. This relates to the justification for the use of LFR.

Specific parameters for its use included discussion on where it would be run, for how long and oversight of this. This was viewed to be of importance to victims of crime, to ensure that they are aware of
situations where LFR may be used. In addition, some queried the equity of service across Scotland and whether it could be rolled out in an equal way.

Throughout the focus groups, participants highlighted challenges in being able to discuss the use of LFR in detail due to the high-level nature of the use cases. If future discussion on LFR is agreed upon, participants
emphasised that further detail and specificity on the parameters of use of LFR would be beneficial.

Some critiqued the use cases presented for discussion. Some argued, particularly in relation to Violence Against Women and Girls (VAWG), that the use cases would not tackle the general occurrence of VAWG, for
example, as this does not tend to occur in public places. These scenarios were thought by others to be broad and covering large public spaces and as such were difficult to discuss in detail.

Parameters of use also considers the use of watchlists. Participants raised concerns regarding the size of watchlists and to ensure these are proportionate and only used in cases of serious crimes. This also related
to transparency of how people may be on a watchlists and understanding how such data is used by police.

Support

There was mixed support across the focus groups and discussions for Police Scotland continuing the conversation and/or implementing LFR at some point in the future.

Across the focus groups and discussions, there was no unqualified support for Police Scotland implementing LFR at some point in the future, with participants expressing a need to see further detail on the matter (e.g.
policy for use) before passing judgement.

Focus Group Organisations

Biometrics and Forensics Ethics Group (BFEG)
Big Brother Watch
British Deaf Association
Crown Office and Procurator Fiscal Service (COPFS)
Convention of Scottish Local Authorities (COSLA)
Coalition for Racial Equalities and Rights (CRER)
Centre for Research into Information, Surveillance and Privacy (CRISP) – University of Stirling
Children and Young People’s Centre for Justice (CYCJ)
Children and Young People’s Commissioner Scotland (CYPCS)
Disability Equality Scotland
Equality and Human Rights Commission (EHRC)
Equality Network
Policing Performance Committee
National Conversation on Live Facial Recognition - Update
His Majesty's Inspectorate of Constabulary in Scotland (HMICS)
Home Office
Information Commissioner’s Office (ICO)
Leverhulme Research Centre for Forensic Science, University of Dundee
Police Investigations and Review Commissioner (PIRC)
Retailers Against Crime
Scottish Biometrics Commissioner
Scottish Government
Sentrysis
Scottish Human Rights Commission (SHRC)
Sikhs in Scotland
SPA Forensic Services
Swinburne University
University of Southampton
Victim Support Scotland