The argument you are having is not whether some form of dataset or intelligence-based tool is a good one. You have to use it; the questions are how you use it, how you assemble it, and how you ensure it is not discriminatory.
… It is really important, because those are the fundamental questions that, if you do not get them right at the beginning of trying to use a modern policing tool, cause it to fall apart.
Source: Matthew Ryder QC, Matrix Chambers
426.In the years since the Stephen Lawrence Inquiry a range of new policing technologies and tools have been developed that were not in existence at the time of the Macpherson report. However, during the course of this inquiry, we have noted the same patterns of racial disparities, and attempts to justify them, in connection with such developments as are evidenced in relation to debates surrounding stop and search. We cover some of these areas briefly in this chapter.
427.The Gangs Violence Matrix (GVM) database was set up by the Metropolitan Police Service in the aftermath of the 2011 riots to serve as an intelligence tool “to identify those at risk of committing, or being a victim of, gang-related violence in London”. However the Mayor of London, Sadiq Khan, has acknowledged that the Gangs Violence Matrix has been controversial in some communities, “resulting in mistrust of the police and rising tensions”.
428.The Metropolitan Police Service told us that it used the Gangs Violence Matrix to “identify those most at risk of gang-related violence, either as offenders or victims”. It added that the Matrix was a vital tool to reduce gang violence, to safeguard those exploited by gangs and to prevent “young lives being lost”. In December 2018 MOPAC completed a review of the Gangs Violence Matrix and made nine recommendations including a “comprehensive overhaul of the Matrix Operating Model”, training for all officers on its use and improvements to the systematic data capture on individuals on the Matrix particularly around needs, referrals and outcomes.
429.The Metropolitan Police confirmed to Amnesty International that as of October 2017 there were 3,806 people on the Matrix. As of 31 March 2019, three months after the 2018 review was published, 3,134 people were listed on the Gangs Violence Matrix. In its Gangs Violence Matrix quarterly figures, as at 31 March 2021, the Metropolitan Police Service reported that the number of individuals on the Matrix had fallen to 2,206; 80% of those on the Matrix were Black and 12% were White.
430.As with stop and search, we heard concerns about the disproportionate number of BME individuals represented on the Gangs Violence Matrix.
431.Liberty expressed concern that the operation of the Matrix as a database of gang membership relied on “crude racial profiling” which fostered a “popular narrative of gang violence which is discriminatory and inaccurate”. While Oliver Feeley-Sprague of Amnesty International UK acknowledged that it was important for the police to have appropriate intelligence-led and risk assessment tools in order to keep everyone safe, he argued that the Matrix did not serve that purpose.
[…] one of the problems is the ill-defined and amorphous way that the gang term is placed and labelled on certain individuals. You hear the word “gang” in all sorts of contexts.
432.Some of the evidence we received also raised concerns about the ambiguity of the Metropolitan Police’s approach to victims and perpetrators of gang-related crime on the Gangs Violence Matrix, including the lack of clarity about whether an individual on the Matrix is recorded as a perpetrator or victim. We were told by Oliver Feeley-Sprague of Amnesty International UK that this lack of distinction, coupled with inadequate police data sharing agreements with partner agencies, could lead the police to treat anyone named on the Matrix as a “risky person, not an at-risk person” and potentially impact upon that individual’s access to housing and other services. He also asserted that despite the police’s argument that the Matrix served as both a preventative and criminal enforcement tool, there was a lack of research or case studies about individuals identified as at risk of violence on the Matrix who had been diverted from violent crime.
433.Deputy Assistant Commissioner Duncan Ball of the Metropolitan Police Service acknowledged the concerns about the distinction between victims and perpetrators and told us that the Gangs Violence Matrix had been amended to address the Information Commissioner’s concerns. He added that it was important to acknowledge the overlap between victims and offenders, illustrating this with an example of an individual who might stab another gang member and subsequently become a victim as a result of that offence.
434.The Metropolitan Police Commissioner told us that the Gangs Violence Matrix was created “to save lives, to keep people safe and to make sure that where possible people are taken away from criminality”. She said the evidence showed that the Matrix helped to protect people, whether as offenders or as victims of serious crime, and ultimately made people “less likely to offend”. However she said that the Matrix had been “interpreted by many in the communities as exactly the opposite of that” which was a concern for the Metropolitan Police. Deputy Assistant Commissioner Duncan Ball cited MOPAC’s statistical findings in its 2018 report on the Matrix, which he said were “compelling in terms of the efficacy” of the Matrix as a tool to address offending behaviour. He said:
[…] six months before an individual is included on the gangs matrix, 42% of the cohort exhibit offending. Once they go on to the matrix, that reduces to 38%. When they come off the matrix, it reduces to 19.8%. That is a downward trend in the offending of gang members”.
435.DAC Duncan Ball argued that if an individual was added to the Matrix the Metropolitan Police could prioritise individuals and apply “the appropriate response—be it a safeguarding response for victims with the local authority, or a police enforcement response” in order to reduce their offending or the chances of re-victimisation as appropriate.
436.However, it is important to note that the 2018 MOPAC review of the Gangs Violence Matrix found that it was not possible to fully understand the reasons for this impact:
It would appear that the Matrix does have a positive impact on reducing levels of offending by and victimisation of the individuals included on it–and that the reductions in these risks are sustained after they have been removed from the Matrix. However, limitations on the data available from partner agencies mean it is not possible to identify the specific reasons for this impact.
437.Giving evidence to this inquiry in July 2019, Oliver Feeley-Sprague said that the “figures on disproportionality” for the tool were “startling” and that the disproportionate representation of young Black men on the Gangs Violence Matrix was “out of all sense of proportion to the demographics or even the relationship between ethnic grouping and crime”; he argued that it showed a lack of police progress since the Macpherson inquiry. By contrast, DAC Duncan Ball told us that while he recognised there was disparity he did not think it was “hugely disproportionate” given that the Metropolitan Police were trying to manage the risk of serious violence in a cohort that was already “severely disproportionately represented as both offenders and victims in the most serious categories of violence”. In follow-up written evidence DAC Duncan Ball referenced the under-25 figures used in MOPAC’s review to evidence a reason for ethnic disproportionality on the Gangs Violence Matrix.
438.Dame Cressida Dick, Commissioner of the Metropolitan Police Service told us that her force had not conducted any polling or consultations with young BME individuals who were disproportionately included on the Gangs Violence Matrix but said that she was keen for her force to consult young people on this matter in the future.
439.The 2018 MOPAC review identified that “the representation of young, Black males on the Matrix was disproportionate to their likelihood of criminality and victimisation” and, recommended that, as part of the required “comprehensive overhaul” of the database operating model, individuals are “added and removed in a standardised, evidence-based manner”.
440.The Mayor of London asserted that one year on from the MOPAC review, the Metropolitan Police had “completed its work and introduced a more evidence-based approach to adding and removing people”. However he committed to further improve transparency and trust in the Matrix by conducting annual reviews “to monitor the demographics and use of data on the Matrix”. He added that MOPAC would publish a progress report on the 2018 review recommendations “in the spring”.
441.In February 2021 the Mayor of London reported that, following MOPAC’s 2018 review of the Gangs Violence Matrix, the overall Matrix population had fallen to its lowest level in seven years, with 2,304 individuals. It also reported that:
442.As part of its update on the Gangs Violence Matrix Review recommendations, it made the commitment that “robust oversight” of the Matrix would be incorporated into a new board which was being created as part of the Mayor’s Action Plan for Transparency, Accountability and Trust in Policing to monitor disproportionality. This board would report to the London Crime Reduction Board, chaired by the Mayor.
443.The police’s use of body-worn cameras was raised by a number of participants in our roundtable with Black and minority ethnic young people from London. Witness E said that the police’s use of body-worn cameras had forced them to be more careful in their interactions with people.
444.HMICFRS told us that police use of body-worn video could provide a “rich source of information” with regard to the extent to which interactions with the public were “appropriate, fair and respectful”. Its 2017 legitimacy inspection found that 35 out of the 43 forces in England and Wales were “either using or piloting the use of body-worn video cameras to record their interactions with the public”. Their report also found that some forces “mandated the use of body-worn video cameras for specific activities, such as when using force or when stopping and searching people” which HMICFRS supported on the basis that it improved the behaviour of the officer and individual and enabled effective scrutiny of the interaction. However another participant [Witness L] told us from his experience that the police sometimes turned off the cameras during their interactions with young people and verbally abused them once the cameras were switched off.
445.Similarly, Dr Rebekah Delsol told us that body-worn cameras had done little to address individuals’ concerns about the use of stop and search. She said that there were ongoing complaints that cameras were being switched off and that police footage was destroyed before people had a chance to make a complaint.
446.In May 2021 the Metropolitan Police Commissioner told us the use of body-worn cameras was “mandatory for all officers for stop and search and use of force incidents”. She said there were a “small number of occasions” when it might not be appropriate or proportionate to record an encounter and that in these cases, “the reason must be recorded and justified”. She added that her force’s stop and searches were subject to scrutiny by its Community Monitoring Group and these groups could view body-worn video as part of their work. In 2019 she confirmed that the Metropolitan Police kept body-worn video footage for 30 days, unless there was “a legitimate, lawful reason not to”. She added that a police officer could request that footage was retained beyond 30 days if there was a justified and legitimate purpose for doing so such as evidence for a court case, or police complaint.
447.In its 2021 report on Disproportionate use of police powers, HMICFRS urged greater use of body-worn video “to learn lessons” but warned that its potential was not being maximised. It stated that “Body-worn video footage, if made available to the public as part of a structured process, could open up policing to wider scrutiny, to help improve encounters and reassure the public”. The report recommended that, by September 2021, forces should:
448.The Commission on Race and Ethnic Disparities supported the recommendation of HMICFRS on body-worn video and went further by recommending that the officer in a case where BWV was not switched on during stop and search should provide a written explanation of the reason, which it said must be available to the individual who was stopped and searched. Written explanations should be reviewed by the supervising officer and action taken through performance or misconduct procedures if the explanation raised concern.
449.In January 2020 the Metropolitan Police announced that it was beginning the operational use of live facial recognition technology (LFRT). Other police forces have trialled the technology in recent years.
450.Concerns have been expressed about the use of LFRT, including about the effectiveness and accuracy of the technology. Nick Glynn told us that that there was a “real risk” that there would be an overreliance on artificial intelligence and that it was not “a magic bullet”. Similarly, Matthew Ryder QC told the Committee that it was “vital to understand the risks of racist practice or discrimination embedding in that kind of policing tool [facial recognition software], which will be the kind of tool in which discrimination manifests itself in future”.
451.The US National Institute of Standards and Technology (NIST) in December 2019 carried out a study that involved 18 million images and assessed the accuracy of face recognition algorithms for demographic groups defined by sex, age, and race or country. NIST found that false positive rates were highest in West and East African and East Asian people, and lowest in Eastern European individuals. NIST noted several potential reasons for this bias, for example it found the quality of the images used was higher in some countries than others and that there were wide differences between the algorithms used: two of the algorithms NIST tested “assigned the wrong gender to black females almost 35% of the time”.
452.The UK’s Centre for Data Ethics and Innovation (CDEI) has reported that human operators may have innate biases and are likely to be better at distinguishing and recognising faces from their own ethnic background than from other backgrounds. The report also noted that an independent review of six Metropolitan Police FRT trials in Romford, Soho and Stratford between 2018 and 2019 found that, of the 42 matches observed by researchers during the trials, 16 were rejected by the human operator as not credible, 4 people were lost in the crowd, 14 were wrongly stopped, and 8 were correctly stopped. The independent review also found that face recognition matches were “verifiably correct” in less than 20% of cases.
453.In September 2019, the High Court ruled that the use of LFRT by South Wales Police was lawful. On 24 February 2020 Dame Cressida Dick told the Royal United Services Institute that “the tech [the MPS] are deploying is proven not to have an ethnic bias”, claiming that “Currently the only bias in it is that it shows it is slightly harder to identify a wanted woman than a wanted man”. In July 2020 she told us that, following the end of its trial period, the technology being used by the MPS “is now equal [in recognition of women] and it was already equal on ethnicity”.
454.In 2020 US software companies including IBM and Microsoft expressed concern about the risks of bias in police use of this technology. Microsoft has called on the US Government to introduce a “national law … grounded in human rights” to govern its use. There have also been calls for new transparency guidelines in the UK. A written answer from the Policing Minister in November 2020 indicated that the Government had no plans to regulate the use of facial recognition technology as evidence.
455.Following the outbreak of SARS-CoV-2 (covid-19) and the implementation of the first national lockdown in March 2020, the Government introduced regulations designed to slow the spread of the disease by restricting people’s freedom to leave home and giving police unprecedented powers to enforce these restrictions, including powers to issue Fixed Penalty Notices (FPNs) to adults they “reasonably believe” have committed an offence under the emergency health regulations. In April 2020, the Committee published its report, Home Office preparedness for covid-19 (coronavirus): Policing which considered enforcement of the new restrictions and made a number of recommendations to the Government.
456.In addition to that report, the Committee sought specific evidence on the enforcement of covid-19 regulations by ethnicity as part of this inquiry, following concerns raised about disproportionality in fines and investigations of individuals from BME communities.
457.On 26 May 2020, the Guardian reported research from Liberty Investigates showing that “Black, Asian and minority ethnic people in England are 54% more likely to be fined under coronavirus rules than White people”. In response to these findings, the National Police Chiefs Council said that it “was not confident that meaningful conclusions” could be made about ethnic proportionality in its data on FPNs issued during lockdown. It said the data was not subject to the same robust quality assurance that takes place with an “established official statistics collection”.
458.On 3 June 2020, the Metropolitan Police published data indicating that “when compared with the composition of the resident population, higher proportions of those in Black and minority ethnic (BAME) groups were overall issued with FPNs or arrested across London as a whole”.
459.Mirren Gidda of Liberty Investigates gave evidence to the Committee on 17 June 2020 where she also raised concerns about data collection:
We need the police forces to drastically improve their collection of ethnicity data, and also relevant other factors such as age. We need to get to a point where we are not having to rely on officer-defined ethnicity, which can often vary from self-defined ethnicity and may not be completely accurate. We need to get to a point where the data that is being published is more transparent.
460.In the same session, Professor Ben Bowling told us that:
The evidence we have so far is that the policing of covid [-19] in the United Kingdom has tended to follow a pattern that is evident in other fields of policing over many years. Black and minority ethnic communities have been disproportionately affected by policing powers. As we have already heard, that is in relation to fines, fixed penalty notices, as well as arrests for covid breaches.
461.In oral evidence to the Committee on 8 July 2020, the Metropolitan Police Commissioner told us that “The first few weeks of the lockdown was the time when we had the vast majority of our enforcement activity”. She said that:
We had only 36 arrests in the whole of London for covid-only legislation in that time period. It was a very small amount of our activity that resulted in enforcement and the vast majority of it was engage, encourage, explain, again and again, to people of all communities. There is disproportionality, absolutely, but it is in such a tiny number and, as I say, the vast proportion of those were people who were being arrested for another offence when there was enforcement. I know it has caused concern. My own view, compared with some other issues, is that it is a lower-order issue.
462.On 27 July 2020, the NPCC published a report, Policing the Pandemic, analysing data relating to covid-19 related FPNs issued between 27 March and 25 May 2020. The actual number of FPNs issued (17,039) represented a rate of 3 fines for every 10,000 residents in England and 6 fines per 10,000 in Wales.
In relation to racial disparity, it found that:
463.The NPCC also warned that the data should be interpreted with caution especially for small forces which had given out few Fixed Penalty Notices or forces that had been affected by people travelling between areas.
464.On the publication of the data, NPCC Chair Martin Hewitt said:
while it is a complex picture, it is a concern to see disparity between White and Black, Asian or ethnic minority people. Each force will be looking at this carefully to assess and mitigate any risks of bias–conscious or unconscious—and to minimise disproportionate impact wherever possible. Many forces have brought in community representatives to help them scrutinise the circumstances around each FPN and if it has been issued fairly.
We are working to develop a plan of action to address issues of inclusion and race equality that still exist in policing… the findings of this analysis will be further considered as part of that work.
465.New policing technologies have developed in the decades since the Macpherson report. These technologies, which clearly could not have been considered by the Stephen Lawrence Inquiry, have given rise to similar kinds of issues about the importance of sustaining confidence among minority ethnic communities, avoiding racial disparities and ensuring fairness in policing. Too often we have seen evidence of new measures or technologies being introduced without sensitivity to the potential impact on race disparities or community confidence. It is vital that police forces, policing institutions and the Home Office have systems in place to ensure that new technology or new measures are implemented fairly, without racial bias and without widening unfair racial disparities.
466.The Metropolitan Police developed the Gangs Violence Matrix as a new way to provide intelligence to tackle serious gang related violence and crime in London, but without robust systems in place to consider racial disproportionality on the database, ensure proper oversight or sustain community confidence. As a result, considerable community concern grew about the use of the database and the high levels of racial disproportionality. The MOPAC review of the database and the commitments since by MOPAC and the Metropolitan Police Service to reform the Gangs Violence Matrix since 2018 are welcome. It is important that MOPAC’s commitment to provide oversight and monitor disproportionality on the Matrix is followed through as part of wider efforts to monitor the potential for racial bias in policing tools as a matter of course.
467.More recently, the serious concerns raised about disproportionality in the use of Fixed Penalty Notices as part of police enforcement of the covid-19 regulations provide cautionary evidence about the need for care and oversight in the way new policing powers are introduced. In the first lockdown Black people were 1.8 times more likely to be subject to covid enforcement measures than White people.
468.Evidence of disproportionality must be carefully considered and presented transparently, with robust systems of independent oversight. Although the NPCC conducted a detailed analysis of the use of covid Fixed Penalty Notices by ethnicity during the first lockdown, in response to issues raised in the media and questions from this Committee, we note with concern that neither the NPCC nor the Home Office have published any further analysis of covid-19 enforcement by ethnicity during subsequent restrictions or lockdowns even though they know there is an unexplained racial disparity. Leaving it to individual forces to follow up is not good enough, especially when the NPCC has pointed out that data analysis is more difficult at local level where smaller numbers are involved. The NPCC and Home Office should be continuing to monitor the data to see whether the racial disparity persists, what the reasons are behind it, and what action may be needed to ensure that there is no unfairness or racial injustice in the use of new powers.
469.New technologies have the potential either to re-build community confidence and/or to badly damage it, depending on the technology, on the way it is introduced and the nature of the oversight. The introduction of police body-worn cameras, if done properly, is a new technology that could help to rebuild community confidence. As we have recommended in the previous chapter, it has an essential role to play in ensuring that stop and search is done fairly under proper oversight. But it is important that body-worn video is used consistently, rather than being left to individual officer discretion. Footage must be provided as part of structured processes of oversight and review, both internally and externally, to facilitate lessons being learned and openness with the public.
470.The police are currently exploring other new technologies such as live facial recognition technology, where serious consideration is needed of the way the technology might apply for different communities and any consequences for racial disparities.
471.As new policing technologies, tools and powers are developed, it is important that there are robust and credible processes in place both to guard against the risk of importing or exacerbating racial disparities and to maximise their potential to demonstrate fairness and build consent in the public.
472.Under the Equality Act 2010 the Home Office and the police have a legal duty to consider the equality impact of new policies, measures or technologies on race equality or other protected characteristics. We do not believe that this responsibility is currently being taken seriously enough. The Home Office, NPCC and College of Policing should work together to identify the range of new policing technologies or measures for which national race equality assessments should be done or where new research and data gathering is needed to anticipate, monitor or swiftly address unjustified race disparities.
715 , The Mayor of London.
717 , Metropolitan Police Service.
718 , Metropolitan Police Service.
719 MOPAC, , December 2018; , January 2021.
720 , Amnesty International p17
721 Metropolitan Police, Gangs Violence Matrix, ; Note that Q4, 2018/19 is the earliest quarterly update available on the Metropolitan Police Gangs Violence Matrix website [accessed 30 April 2021].
722 Metropolitan Police, , as at 31 March 2021.
733 MOPAC, , December 2018, p5.
735 ; .
737 ; .
738 MOPAC, , December 2018
739 Mayor of London, , 16 February 2020.
740 Mayor of London, , 16 February 2020.
741 Metropolitan Police, , created 31 December 2020; Mayor of London for Policing and Crime, , 3 February 2021.
742 Mayor of London for Policing and Crime, , 3 February 2021.
743 Mayor of London for Policing and Crime, , 3 February 2021.
745 , HMICFRS, para.82.
746 , HMICFRS, para.83.
747 , HMICFRS, para.83.
749 Dr Rebekah Delsol, StopWatch.
750 Commissioner of the Metropolitan Police Service.
751 Commissioner of the Metropolitan Police Service.
753 ; .
754 HMICFRS, , February 2021, p18, p20.
755 , March 2021 p22
756 The Guardian, , 24 January 2020.
759 US Department of Commerce National Institute of Standards and Technology December 2019.
760 US Department of Commerce National Institute of Standards and Technology December 2019, p55.
761 Centre for Data Ethics and Innovation 28 May 2020; the Human Rights, Big Data and Technology Project – July 2019.
762 The Human Rights, Big Data and Technology Project – , July 2019.
763 The Human Rights, Big Data and Technology Project, , July 2019.
764 BBC News, , 4 September 2019.
765 The Times, ’, 24 February 2020.
766 8 July 2020
767 BBC News 9 June 2020.
768 The Washington Post 11 June 2020.
769 3 November 2020.
770 House of Commons library briefing, , 9 April 2021.
771 The Guardian, , 26 May 2020.
772 , london.gov.uk [accessed 27 July 2021].
776 National Police Chiefs’ Council, 27 July 2020, p.7.
777 NPCC, , 27 July 2020.
778 National Police Chiefs’ Council, , 27 July 2020.