Use of facial recognition technology by public authorities in the EU

• Author: European Union Agency for Fundamental Rights (EU body or agency)
• Pages: 11-17

FRA Focus

11

5. Use of facial recognition technology by

public authorities in the EU

To date, there is no comprehensive overview on the

use of facial recognition technology in the EU. Many

IT companies offer facial recognition technologies and

there is a strong interest in the use of the technol-

ogy by public administrations for different purposes.

This section focuses on their use for law enforce-

ment purposes.

47

A number of tests of facial recog-

nition technologies were carried out during the past

years by law enforcement authorities in different EU

Member States, although the information available

is limited. Apart from test deployments of live facial

recognition technologies by public authorities in EU

Member States, there is an increased planned use

of facial images in the large-scale EU databases in

the f‌ields of migration and security (see Section 5.2).

Meanwhile, research on the possible use of facial

recognition technologies continues (see Section 5.3).

5.1. Testing facial recognition

technologies by law

enforcement in EU

Member States

FRA interviewed representatives of public authorities

in Germany, France and the United Kingdom about

their possible use and plans of using live facial rec-

ognition technologies for law enforcement purposes.

So far, the police in the United Kingdom has been

most active in experimenting with live facial rec-

ognition technologies. The United Kingdom is the

only EU Member State testing live facial recogni-

tion technologies in the f‌ield with real watchlists.

For example, the South Wales Police has used it at

major events,

48

and the London Metropolitan Police

has carried out several live trials of facial recogni-

tion technologies.

47 To give another example, a Swedish municipality has used

facial recognition technology to monitor the attendance of

pupils in schools. This has led the Swedish Data Protection

Authority to ne the municipality for violating the GDPR.

See European Data Protection Board, “Facial recognition

school renders Sweden’s rst GDPR ne“, 22 August 2019. In

a similar vein, the French data protection authority (CNIL) has

also held that the use of facial recognition technology at the

entrance of two high schools (in Marseilles and in Nice), for

security reasons, appears neither necessary nor proportionate

to the given purpose and violates the GDPR. See CNIL,

“Expérimentation de la reconnaissance faciale dans deux

lycées : la CNIL précise sa position“, 20 October 2019.

48 South Wales Police also tested it for criminal investigation

purposes based on CCTV materials, but retrospectively.

The South Wales Police were the f‌irst to use live

facial recognition technology in the United Kingdom

at large sporting events. The police used it at the

UEFA Champions League f‌inal in June 2017, which

brought about 310,000 people to Cardiff. The tech-

nology was also used at several further events,

including other sports events and music concerts.

Several CCTV cameras were placed at different pre-

selected locations. Depending on the size of the

events, the police constructed watchlists including

several hundreds of people of interest. According to

the independent evaluation report from the trials,

four different watchlists were used for the UEFA

Champions League f‌inal. These include:

a small number of individuals, who were per-

ceived to pose a serious risk to public safety;

individuals with previous convictions for more

serious offense types;

individuals of possible interest to police, whose

presence did not pose any immediate risk or

threat to public safety; and

images of police off‌icers to test the effective-

ness of the system.

The watchlists contained between 400 and 1,200

individuals for the different events. The selection

was based on different possible criteria. However,

no further information on the creation of watchlists

was shared with the evaluators of the trial.49 The

absence of information on how watchlists were cre-

ated makes diff‌icult an assessment of the real pur-

pose, necessity and social need for employing live

facial recognition technology. The f‌irst case on this

issue to come before a court in the European Union

(judgment not f‌inal) arose in a divisional court in

Cardiff. It ruled, in a case directed against the South

Wales Police, that the current national legal regime

is adequate to ensure the appropriate and non-arbi-

trary use of the facial recognition technology called

“AFR Locate”, and that the South Wales Police’s use

to date of “AFR Locate” has been consistent with

49 Davies B., Innes M., and Dawson A., An evaluation of South

Wales Police’s use of Automated Facial Recognition, Cardi

University, September 2018. In addition to these deployments

to locate people, the South Wales Police used FRT to identify

suspects from past crime scenes. Images captured at crime

scenes via CCTV or mobile phone cameras are compared

against a large database of police custody images for

investigation purposes.