Met Police live facial recognition
Released: August 2016
London's Metropolitan Police Service conducted a series of trials of live facial recognition technology across London between August 2016 and February 2019. Trials included the Notting Hill Carnival, Remembrance Sunday, Stratford Westfield shopping centre, and Romford.
The programme was plagued by criticism from civil and privacy rights advocates and technology and legal experts regarding the accuracy of the system, and complaints about inadequate transparency, accountability, and privacy protection.
A July 2019 University of Essex report (pdf) commissioned by the Met Police into its broader use of facial recognition found that only eight of 42 matches were verified as correct, meaning 81% of suspects identified by its system were innocent.
The Met Police responded by saying it was 'extremely disappointed with the negative and unbalanced tone of th[e] report'. As Sky News reported, the force 'prefers to measure accuracy by comparing successful and unsuccessful matches with the total number of faces processed by the facial recognition system. According to this metric, the error rate was 0.1%.'
Human rights law compliance
Legal and ethical standards
A report (pdf) published in October 2022 by University of Cambridges Minderoo Centre for Technology and Democracy researchers found that the Met Police's trials suffered from inadequate transparency and accountability, poor privacy, and failed to meet minimum expected ethical and legal standards.
The researchers went on to argue that live facial recognition technology should be banned from use in streets, airports and any public spaces in the UK.
Resumption of monitoring
In April 2023, the Met Police announced it was to resume all forms of facial recognition on the basis of a study it and South Wales Police had commissioned the UK National Physical Laboratory to carry out.
The study found that NEC's NeoFace system showed 'no statistically significant race and gender bias,' and argues that facial recognition surveillance can protect human rights as it reduces and prevents crime.
Investigations, assessments, audits
Science & Technology in Policing (2023). Operational Testing of Facial Recognition Technology
Minderoo Centre for Technology & Democracy (2022). A Socio-Technical Audit: Assessing Police Use of Facial Recognition (pdf)
Big Brother Watch (2020). Briefing on facial recognition surveillance (pdf)