User menu

Is there a legitimate role for facial recognition in policing and law enforcement?

This is the testimony given by Chief Constable Jeremy Vaughan ahead of the discussion on facial recognition technology held at the London School of Economics on the 14 June 2022.

 

“I welcome the opportunity to attend here this evening, to contribute to the debate on whether there is a legitimate role for facial recognition in policing and law enforcement. I think that there is, and this is a view which seems to be shared by the majority of the public.

The police have a duty to prevent crime and disorder and protect the public, but we do so with the approval and consent of the public. Therefore, in our application of this duty, the actions of policing must be legitimate – that is to say, whatever policing does, must be proportionate, lawful and necessary.

The balance between people’s freedoms and rights and police’s use of power (and application of new technologies to facilitate that use) must continue to be at the forefront of our minds at all times. The duty here is on policing to operate within the law, and that the law is sufficient, is a debate which must continue.

The criminal landscape is becoming ever more complex and sophisticated, and law enforcement must innovate so that we can respond to this complexity and protect those who need us the most. In this digital era, technology has created a new and emerging frontline of crime.

In using Facial Recognition Technology or FRT, it is always a police officer who will make the decision, drawing on experience and knowledge. It is important that any new technologies do not fetter individual officer discretion to make a decision in any given set of circumstances.

Overview of FRT

Policing has always spent a lot of time and public money trying to identify and locate offenders and vulnerable persons. In many cases hundreds of hours are spent looking for individuals with no success.

Not all use of FRT in law enforcement is the same.  Policing currently uses the technology in three distinct ways:

  1. Retrospectively – where an image obtained post an incident is compared to other images held on the custody image database.
  1. Live Facial Recognition is used when cameras are overtly placed in public places supported by an intelligence case. The software receives the live feed and compares faces against a watchlist in order to locate individuals. After which point, both the image and biometric data are immediately deleted.
  1. Operator Initiated Facial Recognition which is currently being used by police officers on their mobile phones as part of a trial. To date, this has involved sixty officers from a variety of policing backgrounds over a three-month period.

In each case it is the police officer or staff member, who makes the ultimate decision on how to use the information. 

Use Cases

South Wales Police has utilised Retrospective Facial Recognition since 2017, against images lawfully held on our custody image system.

Prior to the retrospective application officers would circulate images both internally and externally taking on average at least ten days to make an identification, if indeed one was made at all. It was heavily reliant on an officer seeing an image and recognising an offender.

Now, when we have an image of a suspect, we will utilise the technology to look for a match with one of the images from our custody database. More often than not matches are made on the same day of submission, often within minutes.

The efficiency savings within the identification process alone are huge, but more importantly, they improve our ability to protect victims and bring offenders to justice.

Since using the technology, South Wales Police has achieved nearly 4,000 possible matches, averaging 100 every month.

It has been successfully used for all offence types, from murder to the more minor offences. The technology has significantly assisted South Wales Police bring offenders to justice, and in turn help protect victims of crime.

For example, in November 2019, a woman reported that she had been attacked on a bus in Cardiff. Her screams raised the alarm, and the suspect was chased off.

Using a grainy image captured by CCTV on a bus, the technology was able to assist in positively identifying the individual against a 14-year-old police custody image.

As a result, a male, who had already spent half his life in jail for attacks on women, was sentenced to two life terms, to run concurrently.

Without Retrospective Facial Recognition, the CCTV image would have been passed to local officers, then more widely within the force, perhaps even involving an appeal to the public, to help identify a suspect. We may never have found him as he hadn’t been in our custody for 14 years, therefore many police officers working locally would have moved on or even retired from policing.

Live

Live Facial Recognition Technology has been deployed on over 70 occasions to include deployment at major events, pop concerts and crime operations.

This has culminated in 75 persons being identified and arrested for offences including rape and other serious offences.

People Rights – tested, scrutiny and challenge

I am acutely aware that the technology has already been adopted by broader society and the private sector. I do not comment on its use outside of policing, as this is a matter for others. 

It is right though that law enforcement is subject to the highest level of scrutiny. After all none of us can choose which police force to use, that choice is made for us and therefore we should be subject to extraordinary scrutiny.

To that end I welcome the scrutiny provided by existing national Regulators and Commissioners along with the local scrutiny being provided by Police and Crime Commissioners.

South Wales Police has been subject to two legal proceedings regards the use of Live Facial Recognition: the first in the Divisional Court, and second at the Court of Appeal.

Whilst the Court of Appeal recognised the legal basis for policing to overtly deploy Live Facial Recognition, they also instructed that South Wales Police needed to do more regards ‘who’ could be put in a watchlist and ‘where’ the technology could be used. The Court held that the police had too much discretion in this area.

The Court went on to state that the Surveillance Camera Commissioner (SCC) Code does deal with such matters and in principle could deal with specificity with what the requirements are for inclusion of a police force’s watchlist. In effect, the Court signposted the way for there to be a national Code of Practise and /or local policies to remedy the issues found by the Court.

The SCC Code was amended in August 2021 adopting elements of the Court of Appeal judgment as have our policy documents.  These are available for public consumption and located on our website.

The College of Policing as the professional body for policing has also since published Authorised Professional Practise (APP) for the Overt Use of Live Facial Recognition, this again is based on the Court judgment and the (amended) Surveillance Camera Commissioner (SCC) Code.

Taken together these are important steps in providing approved practice and guidance into a consolidated online format ensuring a consistent and compatible approach to all police forces on a national basis.

Furthermore, whilst the Court recognised that there was no clear evidence that Live Facial Recognition was biased on the grounds of race or sex, they required the police to do everything they reasonably could, in fulfilling their Public Sector Equality Duty, to understand potential biases.

Policing is addressing this in three distinct ways. Firstly, using studies that have been carried out by the National Institute of Standards and Technology. 

Secondly, seeking clarity from suppliers regards their demographic differential performance.

Lastly and of most significance, policing has commissioned the largest independent academic operational evaluation regards equitability focusing on age, gender, and ethnicity across the three identified facial recognition use cases. This study is currently underway and is being independently led by the National Physical Laboratory with the results to be available in October 2022.

Democratic Society

Having been subject to and having positively engaged in two court processes, it is my view that we have a duty to push on with the application of the technology that will assist to protect the public.

As the technology changes and adapts there may be a need to revisit the legal landscape which policing operates within. We live in a democratic society, as society views change, this will need to be continually considered by lawmakers and in future there may indeed be a role for a different legislative framework, different checks and balances, and new guidance.

Public Support

There is no doubt we face a significant challenge in continuing to explain the way in which policing utilises Facial Recognition Technology.

It is the experience from South Wales Police that when we effectively explain how policing uses the technology, the public are often very supportive, even at times frustrated that we do not apply the technology far more broadly. 

It is also important to point out that South Wales Police has not received any complaints or made any false arrests regards use of the technology.

An Ada Lovelace Institute study stated that 71% agreed with the statement “the police should be able to use facial recognition on public spaces, provided it helps reduce crime”.

‘The Information Commissioner’s Opinion published June 2021 also includes a section on public perception which showed that 82% of respondents indicated that they found it acceptable for the police to deploy the technology.

Summary

I am absolutely convinced of the benefit Facial Recognition Technology has had in policing in assisting to locate and identify offenders as well as protecting those that are most vulnerable. I think moving forward the technology in all use cases could have a greater role to play in protecting people, and therefore I believe that there is a legitimate role for facial recognition in policing and law enforcement."