UK facial recognition system shows bias against Black, Asian people and women
Official testing by the UK Home Office found that a facial recognition system used by police forces exhibits significantly higher false positive rates for Black people, Asian people and women compared to white men. Police forces previously lobbied to reverse safeguard measures that had reduced the bias, citing reduced investigative leads.
What happened
Testing by the National Physical Laboratory found the UK police facial recognition system, used over 25,000 times monthly, incorrectly flagged Black people as matches 5.5 per cent of the time, compared to 0.04 per cent for white people, and Asian people 4 per cent of the time. For women the false positive rate was 3.8 per cent versus 1.5 per cent for men, with Black women showing the highest false positive rate at 9.9 per cent.
The Home Office acknowledged the bias in December 2025. Documents revealed that police forces had lobbied to reverse a higher confidence threshold that had been ordered in October 2024 after the NPL found bias, because the threshold reduced investigative leads from 56 per cent to 14 per cent of searches.
Reported impact
The civil rights group Liberty estimated that thousands of Black and Asian people had been wrongly flagged during the seven years facial recognition technology has been used in Britain. The Association of Police and Crime Commissioners said the lack of harm so far was 'more by luck than design'.
Index analysis
The system correctly identified Asian suspects 98 per cent of the time, falling to 91 per cent for white suspects and 87 per cent for Black suspects. The algorithm also performed worse for people under 41, with a 5.2 per cent false positive rate for people in their 20s compared to 0.1 per cent for those over 41.
The NPCC documents showed that increasing the confidence threshold significantly reduced bias across race, age and gender but reduced operational effectiveness. At certain settings the algorithm produced false positives for Black women almost 100 times more frequently than white women.
Governance controls
The Home Office said a new algorithm had been developed in response to the findings. The government opened a ten-week consultation on plans to widen the use of facial recognition technology, with policing minister Sarah Jones describing it as 'the biggest breakthrough for catching criminals since DNA matching'.
The Association of Police and Crime Commissioners said the technology had been 'deployed into operational policing without adequate safeguards in place' and criticised ministers for not making the problems public sooner.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.