Expert challenges Met Police's bias-free live facial recognition use claims

Claims made by the Metropolitan Police Service (MPS) that its use of live facial recognition (LFR) technology is bias-free have been challenged by Professor Pete Fussey, a leading expert who conducted the only independent academic review of LFR. Despite MPS claims of reforms following a 2023 study from the National Physical Laboratory (NPL), Fussey argues that the report does not adequately support the police's statements regarding bias. 

The NPL study examined the sensitivity settings of the LFR system, noting that a setting of 0.56 exhibited bias, while a setting of 0.6 resulted in seven false positives, all from individuals belonging to ethnic minorities. The study involved 178,000 images and tested 400 volunteers in crowds across London and Cardiff over 34.5 hours. Although the report suggested no statistically significant bias at settings of 0.6 or higher, Fussey contends that this conclusion is drawn from an inadequate sample size. He criticises the MPS for extrapolating universally applicable conclusions from such limited data. The current LFR sensitivity is set at 0.64, which the MPS claims produces no false matches. However, Fussey emphasises that proper assessments of bias were not conducted at this setting, raising concerns about the claims of bias elimination.