Mass Surveillance Fails: Manchester Police Facial Scans Lead to Minimal Arrests

On a sunny August weekend, the Manchester Caribbean Carnival, a tradition since 1971, drew tens of thousands of revellers to Moss Side and Hulme. Amidst the vibrant music, colourful costumes, and celebrations of heritage, an unexpected sign near a police presence informed attendees that Live Facial Recognition (LFR) technology was being deployed. Every carnival-goer entering the event had their face scanned and checked against a police watchlist, a measure Greater Manchester Police (GMP) stated was for 'an extra layer of security'.

GMP reported that the LFR deployment at the carnival generated 34 alerts, leading to 11 arrests for various offences, including breach of a sexual harm prevention order, fraud, and theft. Among the alerts were also 11 registered sex offenders and five individuals known to carry knives, in addition to the identification of a missing child. For the police, these figures validate the technology's effectiveness, especially considering past violent incidents, such as stabbings and shootings, at post-carnival parties.

However, critics, including Jasleen Chaggar from Big Brother Watch, challenge the ethical implications, questioning the extent of acceptable surveillance in public spaces, particularly at a cultural event where the vast majority of attendees are innocent. Chaggar condemned the targeting of an event that originated as an act of resistance and freedom with 'Orwellian' technology, arguing that law-abiding citizens should not be subjected to a 'digital police line-up'.

The debate has extended to Parliament, with Rebecca Long-Bailey, Labour MP for Salford and Eccles, joining a cross-party coalition of parliamentarians and civil rights groups demanding a halt to facial recognition surveillance by UK police and private companies. Signatories argue that such mass scanning fundamentally alters the citizen-state relationship, treating the public as a 'constant police line-up' without sufficient democratic oversight. Unlike the European Union, which has strict, narrow exceptions for real-time public facial recognition, the UK lacks specific laws or regulations governing its use, leading to concerns about 'sleepwalking into a surveillance state'.

Since October 2025, GMP has deployed LFR 101 times across Greater Manchester, scanning over 1.3 million faces, resulting in 126 arrests and at least 53 subsequent charges. The force's Home Office-funded LFR vans, equipped with fixed cameras and biometric software (NeoFace algorithm by NEC Corporation), have been used at high-profile events like the Brit Awards, MOBOs, football matches, and town centre operations, including an operation in Piccadilly Gardens to find wanted criminals.

When a person passes by, the LFR software analyses facial features and cross-references live feeds against a watchlist. If a potential match is found, an alert is generated for a police officer to assess and decide on further action, ensuring arrests are 'necessary, proportionate, and lawful'. GMP states that images with alerts are deleted immediately or within 24 hours, and biometric data of those who do not trigger an alert is 'automatically and immediately deleted'. The force also advertises its use in advance and supports it with an Equality Impact Assessment (EIA).

Despite GMP's assurances of rigorous testing and oversight, an audit by the Information Commissioner's Office (ICO) in May identified gaps in governance, including unrecorded staff use of the system and generic user accounts, creating risks of undetected inappropriate access. The ICO gave GMP's LFR operation a 'reasonable' assurance rating, but noted 'considerable scope for improvement' in its May 2026 audit of retrospective facial recognition (RFR).

RFR, which involves searching captured images (CCTV, body-worn footage) against databases, has faced persistent questions regarding demographic bias. Home Office-commissioned research found significantly higher false-positive rates for Black (5.5%) and Asian (4%) faces compared to White faces (0.04%). This bias can lead to real-world consequences, as seen when Thames Valley Police mistakenly arrested Alvi Choudhury for burglary due to a faulty RFR match.

The legal history surrounding police use of facial recognition includes a landmark 2020 case where Liberty successfully argued that South Wales Police's (SWP) use of automated LFR violated privacy, data protection, and equality rights, though SWP later resumed use. Ruth Ehrlich, interim director of external relations at Liberty, emphasized that the rapid expansion of LFR without a clear legal framework constitutes a serious intrusion into privacy and risks undermining fundamental rights and freedoms.

While the Manchester Caribbean Carnival is typically peaceful, the policing rationale for increased security stems from serious crimes at after-parties, such as the 2022 fatal stabbing of Javell Morgan. However, critics question the selective deployment of LFR, noting its absence at other large events like Parklife that also record numerous arrests. Many carnival-goers expressed feelings of invasion of privacy and fear, highlighting the 'dehumanising' aspect of treating community members as potential suspects. GMP maintains that LFR helps identify wanted offenders and vulnerable missing people, but a Freedom of Information request from the Manchester Evening News for detailed breakdowns of alerts and accuracy data was met with a response stating the information was not easily retrievable.

The fundamental debate surrounding facial recognition has evolved beyond its technical efficacy to questions of where it should be used, whose faces should be scanned, what safeguards apply, and, crucially, who makes these decisions. As Jasleen Chaggar compellingly argued, the idea of remote scanning of faces is no more reassuring than being fingerprinted on the high street 'just in case', underscoring the deep discomfort many feel about pervasive surveillance.