Met defends facial recognition plan for Carnival

A red Metropolitan Police van with a large sign reading “Live Facial Recognition In Operation” is parked on a busy street, as pedestrians walk past in the sunshine.Image source, Getty Images
Image caption,

Sir Mark Rowley said Live Facial Recognition (LFR) technology can help keep people safe at Notting Hill Carnival

  • Published

The Met Police commissioner has defended Live Facial Recognition (LFR) technology as a tool that helps officers locate people on watch lists, ahead of its use at Notting Hill Carnival.

Sir Mark Rowley responded to calls from civil liberty and anti-racism groups to drop its use of LFR, which treats people "as potential suspects", at the bank holiday event.

"Our aim is to ensure that Carnival remains a safe and inclusive celebration for all. It is our operational judgement that LFR has an important role in delivering on this aim," Sir Mark said in a letter.

He added that almost 350 arrests were made at Carnival last year for a range of offences including homicide, rape and possession of weapons.

On Sunday, 11 organisations penned a letter to Sir Mark, where they described LFR as "a mass surveillance tool that treats all Carnival-goers as potential suspects and has no place at one of London's biggest cultural celebrations".

They said the decision to reintroduce the technology at Carnival was "deeply disappointing" and argued it could be "less accurate for women and people of colour".

'Learning from experience'

Sir Mark said the force had designed an extensive and complex policing operation to keep carnival-goers, expected to be more than one million in number, safe.

He wrote: "We acknowledge that when LFR was previously deployed at Notting Hill Carnival in 2016 and 2017, it did not build public confidence.

"At that time, the technology was in its early stages and the algorithm's performance was limited. The legal and oversight position was also very different."

He said that since, the technology had made "considerable progress", and had an improved algorithm that performed at a higher standard.

"We have also refined our operational approach - including not using LFR within the Carnival footprint," he added.

"These developments reflect our commitment to learning from experience and improving how we support public safety while maintaining trust."

A general view of crowds of people on Ladbroke Grove, London, during the Notting Hill Carnival celebration. Image source, PA Media
Image caption,

Sir Mark said safeguards were in place to ensure the force used the technology in a non-discriminatory way

The groups concerned about the technology included Liberty, Big Brother Watch and the Runnymede Trust. They highlighted an ongoing judicial review brought by Shaun Thompson, a black Londoner who says he was wrongly identified by the system and detained.

Their letter stated: "Notting Hill Carnival is an event that specifically celebrates the British African Caribbean community, yet the [Metropolitan Police] is choosing to use a technology with a well-documented history of inaccurate outcomes and racial bias."

Sir Mark said the force had selected an LFR algorithm with care, which "does not perform in a way which exhibits bias".

He added that safeguards were in place to ensure the force used the technology in a non-discriminatory way.

Referencing the number of arrests last year, Sir Mark said that a small minority of Carnival attendees "have used the event to commit serious crimes".

"Our use of LFR is part of a much broader strategy to locate, disrupt and deter the minority who pose such risks," he said.

He concluded: "Where we know that LFR can help locate individuals the police need to speak to, and those people pose a public safety risk to the many seeking to enjoy Carnival, it is entirely reasonable to ask - why wouldn't we use it in this context?"

Listen to the best of BBC Radio London on Sounds and follow BBC London on Facebook, external, X, external and Instagram, external. Send your story ideas to hello.bbclondon@bbc.co.uk, external