Facial recognition cameras have scanned the faces of more than 125,000 people in London in a little over a year, but resulted in just nine arrrests, according to figures obtained by the Standard.
Five live facial recognition cameras used in Oxford Circus from February 27, 2020 to July 16, 2022, resulted in an estimated 125,786 faces scanned by the Met Police, according to their own figures.
Despite the large number of faces scanned by the cameras, only nine arrests were made over the course of the five operations.
Human rights groups fear the system has not been adequately tested to ensure its efficacy and may violate the privacy of members of the public.
“We should all be able to go about our lives without the threat of being watched, tracked and monitored by the police,” Sam Grant, head of policy and campaigns at human rights group Liberty, said.
“Facial recognition technology is a discriminatory and oppressive surveillance tool that completely undermines this ideal. It, in effect, takes and uses our personal biometric data without our knowledge or consent,”
“We are deeply concerned to see the continued expansion of mass surveillance tools that have no place in a rights-respecting democracy.”
Facial recognition cameras work by analysing facial features to generate a mathematical representation. This representation is then compared against known faces in a database in order to identify possible matches against persons of interest.
Of the five operations carried out by the Met, a deployment of a camera on January 28, 2022 drew a watchlist of 9,756 but resulted in just four arrests from an estimated 12,120 faces scanned. Of the 9,756, 11 were deemed to be alerts from the watch list.
Over the course of three operations on July 7, 14 and 16, 105,066 faces were scanned but only four people were arrested, the data showed. The data did not specify what the arrests were for.
Over that time, 20,159 people were on the watchlist, and eight alerts were triggered.
The Met warns the public that facial recognition software is in use and says any biometric data which does not trigger an alert, is deleted immediately.
It says data held on the encrypted USB memory stick used to import the watch list, is deleted as soon as practicable, and always within 24 hours following the conclusion of the deployment.
Any match which triggers an alert on the system, however, is deleted as soon as practicable and in most cases within 31 days.
Overall, the Met says the technology is only used to “locate dangerous individuals and those who pose a serious risk to our communities”.
“Live Facial Recognition (LFR) is a crime-fighting technology the Met uses to locate dangerous individuals and those who pose a serious risk to our communities,” a spokesman said.
“Operational deployments of LFR technology have resulted in arrests for serious offences including conspiracy to supply Class A drugs, assault on emergency service workers, possession with intent to supply Class A & B drugs, grievous bodily harm and being unlawfully at large having escaped from prison.
“Many people passing the LFR system are of no interest to the Met and their biometric data is automatically deleted. However, all those passing the system are safer because of the intelligence-led arrests made.
“The very low false alert rates across our operational deployments of between 0% and 0.08% confirm that we are focusing our technology on those wanted for serious offences, helping the vast majority of the public to enjoy London’s West-End.”
The Met says it has been continually updating its algorithm recognising its “commitment to use technology which is both accurate and effective”.