Civil liberties campaigners have said a proposal made by Keir Starmer on Thursday to expand the use of live facial recognition technology would amount to effectively introducing a national ID card system based on people’s faces.
Silkie Carlo, director of Big Brother Watch, said it was ironic that the new prime minister was suggesting greater use of facial recognition on the same day that an EU-wide law largely banning real-time surveillance technology came into force.
“The expansion of live facial recognition means millions of innocent Britons will be subject to automated identity checks,” Carlo said. “These are the surveillance tactics of China and Russia and Starmer appears to ignore the implications for civil liberties.”
So far, live facial recognition has been widely used in the UK by the Metropolitan Police and South Wales Police as a real-time aid to help officers detect and prevent crime, including at public events such as last year’s coronation. It is capable of scanning more than 100 faces per minute for identification.
But on Thursday, Starmer suggested it should be rolled out more widely, in response to violent riots across the country following the stabbing and murder of three girls in Southport, Lancashire.
Pledging to create a national policing capability to tackle the unrest, the new prime minister said forces needed to work better together, sharing intelligence and engaging in a “wider deployment of facial recognition technology”.
Details were scant, but immediately afterwards Starmer suggested that rioters could be subject to “criminal behaviour orders to restrict their movements before they can even board a train”, implying wider use of live facial recognition at transport hubs such as train stations.
Daragh Murray, a senior lecturer at Queen Mary University of London, said: “There is a clear danger that in responding to tragedy and public unrest we will expand and consolidate policing without proper scrutiny. Given that police have responded to riots and disturbances for decades, why is facial recognition needed now?”
In the UK there is no specific law regulating the use of facial recognition software and as a result its implementation is largely defined by police forces, who use it to track missing people and criminals, but also, The Met says“people who are on a ‘watch list’ and are wanted by the police.”
The Met policy statement The watch list includes people “wanted by the courts,” people subject to bail or judicial restrictions, but also people “about whom there are reasonable grounds to suspect that the individual depicted is about to commit a crime” in the future.
Campaigners say such broad definitions could allow police to target legitimate protesters, not just people with violent intentions. Facial recognition has been used at the British Grand Prix for the past two years, scanning hundreds of thousands of people, after Just Stop Oil protesters took to the track during Formula One’s first round of 2022.
There have also been concerns about the accuracy of live facial recognition, with black people more likely to be misidentified than other racial groups. The Met says it can re-tune the facial recognition algorithm to remove demographic discrepancies, but Big Brother Watch said doing so would make the algorithm less effective overall.
In a 2020 ruling, a court found that South Wales Police had failed to adequately investigate whether the software had any racial or gender bias. Police say there has been a “substantial improvement” in its accuracy, and research commissioned by the Met suggests the chance of a false match is now one in 6,000.
Big Brother Watch is launching a legal challenge to the Met’s use of the technology alongside Shaun Thompson, a community worker who tackles knife crime, who says he was misidentified and wrongly stopped at London Bridge station in February by officers who demanded he give his fingerprints to prove who he said he was.
Earlier this year, Met officials described the technology as a “paradigm shift” and said that when deployed, it helped make one arrest every two hours.