The UK Home Office has acknowledged that facial recognition technology used by police is more likely to produce false matches for Black and Asian people than for white people, and has launched a public consultation on a new legal framework for law enforcement use of biometrics and facial recognition in England and Wales. The move comes as the government continues to support a wider rollout of live and retrospective facial recognition in policing, but faces growing scrutiny over bias, transparency, and the impact on civil liberties.
New testing by the UK’s National Physical Laboratory (NPL), commissioned by the Home Office, found that one algorithm currently used with the Police National Database produced false positive rates of around 4.0 percent for Asian subjects and 5.5 percent for Black subjects, compared with 0.04 percent for white subjects, with the error rate rising to nearly 10 percent for Black women. The Home Office has said it has since worked with the vendor on a revised algorithm that performs more evenly, but the figures have intensified concerns that deployments could disproportionately affect minority communities, especially when used in live deployments and at scale.
In parallel, the Home Office has launched a ten-week consultation on a new legal framework covering police use of biometrics, facial recognition, and similar technologies. The consultation paper seeks views on issues including when and how live and retrospective facial recognition should be used, what sources police should be allowed to draw watchlists from, retention and governance of biometric data, and the roles of oversight bodies such as the Biometrics and Surveillance Camera Commissioner and the Information Commissioner’s Office. It also asks how existing legislation should be updated to account for emerging technologies and cross-force deployments. A new factsheet and public attitudes survey published alongside the consultation emphasize broad public support for some uses of the technology but also show ongoing concerns about privacy, misuse, and false identification.
The current debate sits against the background of earlier Home Office-commissioned testing that had been used to argue that the Metropolitan Police’s NeoFace-based live facial recognition system was effectively free of demographic bias. In a 2023 evaluation, NPL reported that the NeoFace V4 algorithm could be operated at certain similarity thresholds where there was no statistically significant difference in false positive rates between demographic groups, and police forces and ministers repeatedly cited this to claim that the system had no significant demographic variation. However, the same and subsequent testing also showed that at other operational settings the algorithm produced higher false positive rates for Black subjects, and civil liberties groups have continued to challenge the accuracy and legality of deployments.
The consultation raises questions that extend beyond UK policing. It highlights how vendor-quoted bias-free results often depend on specific operating thresholds, data sets, and use cases, and how national-scale deployments of biometric systems will increasingly be tied to formal legal frameworks and public accountability processes. Vendors and relying parties in the facial recognition and broader biometric ecosystem should expect greater demands for independent testing, transparent reporting on demographic performance, and governance models that can withstand regulatory and legal scrutiny when systems are used for high-stakes identification.
Sources: The Guardian, The Times, UK Government, UK Government, UK Government, Liberty, NPL
—
By Cass Kennedy
