Skip to content
Submit incident
AII-1002-11689
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

Met Police Deploys Live Facial Recognition at Notting Hill Carnival Despite Bias Concerns

The Metropolitan Police deployed live facial recognition at Notting Hill carnival despite concerns about bias and an Equality and Human Rights Commission finding that such police use is unlawful. An independent expert challenged the force's claim that a recent study showed the system was bias-free, citing insufficient sample size and documented false positives among ethnic minorities.

23 August 2025 Β· Metropolitan Police Service Β· Discrimination and fairness
23 August 2025
AII-1002-116892 sources ↓
Quick facts
Category
Discrimination and fairness
Country
United Kingdom
Source
The Guardian
Date
23 August 2025
Drafted by
Raidu Incident Detection

What happened

The Metropolitan Police planned to deploy live facial recognition at Notting Hill carnival in west London at two sites on the approaches to the event. The force insisted on using the technology despite the Equality and Human Rights Commission saying police use of LFR is unlawful.

The Met claims to have reformed its use of live facial recognition following a 2023 study commissioned from the National Physical Laboratory and states the system is now bias-free. However, Prof Pete Fussey, who led the only independent academic review of police use of facial recognition, said the Met's claims about the absence of bias are not substantiated by the facts in the NPL report.

Reported impact

The NPL study found bias at a setting of 0.56 sensitivity. At a setting of 0.6, the system produced seven false positives, all of which were people from ethnic minorities flagged as wanted.

The Equality and Human Rights Commission has stated that police use of live facial recognition is unlawful.

Index analysis

The NPL study tested the system after 178,000 images were entered, with 400 volunteers providing approximately 4,000 recognition opportunities across crowds estimated at more than 130,000 people at five sites in London and Cardiff. Testing took place in sunny conditions for 34.5 hours, which Fussey said is shorter than testing in some other countries.

Fussey argued the sample size was too small to support the Met's claims of no statistically significant bias at a sensitivity setting of 0.6 or higher.

Governance controls

The Metropolitan Police commissioned a study from the National Physical Laboratory to test the system for bias and adjust sensitivity settings accordingly.

The Equality and Human Rights Commission has determined that police use of live facial recognition is unlawful.

Classification

Category
Discrimination and fairness
Industry
Government, security and defence
Country
United Kingdom
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Human or fundamental rights
Harmed
General public
Failure modes
Biased model, Misidentification
Consequences
None recorded
AI systems
Metropolitan Police Live Facial Recognition System
Regulations
GDPR (cited)

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

The Guardian
Expert rejects Met police claim that study backs bias-free live facial recognition use
23 August 2025
Biometric Update
Police claims on live facial recognition bias not supported by facts: UK professor | Biometric Update
25 August 2025