Reevaluating Live Facial Recognition in UK Policing: Implications and Challenges

Live Facial Recognition (LFR) technology is increasingly integrated into UK police operations, subtly transforming how suspicion is formed and acted upon. This shift raises critical questions about the effectiveness of traditional safeguards, particularly the role of human judgment in policing decisions.

Expansion and Routine Use of LFR in Policing

Once a sporadic tool deployed only occasionally, LFR has become a staple in law enforcement across major UK cities such as London and Cardiff. Annually, millions of faces are scanned by police using this technology, marking a significant escalation from its initial limited use. Facial recognition cameras are now a common presence at public events and in crowded urban spaces, reflecting a broader trend toward surveillance-driven policing.

Human Oversight: Myth or Reality?

Police authorities maintain that human officers ultimately decide whether to engage individuals flagged by LFR systems, aiming to prevent wrongful stops caused by false positives. However, emerging research suggests that this “human-in-the-loop” safeguard is compromised. Instead of independent judgment, officers often act as intermediaries who interpret and respond to algorithm-generated alerts, which can prime them to view flagged individuals as inherently suspicious.

How Technology Shapes Police Perception

Studies by sociologists Pete Fussey, Bethan Davies, and Martin Innes highlight that LFR is not merely a neutral tool but a socio-technical system that both influences and is influenced by police discretion. The initial identification of suspects shifts from officer intuition to computer-generated suggestions, altering the dynamics of suspicion and decision-making on the street.

Further analysis by Karen Yeung and Wenlong Li in 2025 emphasizes that an LFR match alert alone does not meet the legal threshold of “reasonable suspicion” required for police to lawfully stop and question individuals in England and Wales. Despite this, officers often proceed with stops based solely on these alerts, creating a legal and ethical gray area.

Real-World Consequences: Cases and Public Concerns

Instances of wrongful identification and subsequent police interactions have sparked public outcry. For example, in early 2025, Shaun Thompson, an anti-knife crime activist volunteering in Croydon, was mistakenly identified by the Metropolitan Police’s LFR system. Despite presenting valid identification, he was subjected to repeated demands for fingerprint scans and threats of arrest, illustrating how LFR can lead to invasive and unjustified police encounters.

Such incidents have prompted calls for judicial reviews and stricter oversight. Thompson’s case is set to be examined in a judicial inquiry scheduled for January 2026, aiming to prevent similar injustices.

Moreover, even without generating alerts, LFR deployments can provoke confrontations. In 2019, individuals covering their faces near an LFR vehicle in Romford were stopped by police, highlighting public unease about biometric data collection and privacy.

Surveillance and Suspicion: The Broader Societal Impact

Experts warn that LFR reverses traditional surveillance logic by treating everyone passing through camera fields as potentially suspicious from the outset. This “presumption of suspicion” can lead to automation bias, where officers disproportionately rely on technology outputs over their own judgment, effectively turning human decisions into automated ones.

Fussey, Murray, and criminologist Amy Stevens advocate for robust monitoring systems to track when and how officers diverge from or comply with algorithmic recommendations, ensuring that technology remains an advisory tool rather than a directive force.

Watchlist Formation: The Roots of Bureaucratic Suspicion

Central to LFR’s operation is the creation of watchlists, which determine who is subject to surveillance. These lists are primarily populated with custody images held by police, focusing attention on individuals already within the criminal justice system. This practice institutionalizes a “bureaucratic doubt,” where suspicion is generalized rather than incident-specific.

Research reveals that watchlists disproportionately include young people and ethnic minorities, particularly those of African Caribbean descent, compounding existing social biases. Additionally, technical limitations of facial recognition-such as reduced accuracy for older adults, women, and people of color-further complicate fair application.

Legal scholars Karen Yeung and Wenlong Li have raised concerns about the vague and expansive criteria used to compile watchlists. Contrary to police assertions that only serious offenders are targeted, watchlists often contain individuals linked to minor offenses like shoplifting or drug possession, as well as people barred from public events or classified as vulnerable.

In 2023, senior police officials acknowledged to a parliamentary committee that LFR operates under a “bureaucratic suspect” framework, where generic crime categories, rather than individualized threat assessments, guide watchlist inclusion. This ambiguity has led to legal challenges, including a 2020 Court of Appeal ruling that deemed South Wales Police’s LFR use unlawful due to overly broad watchlist criteria.

Ongoing Issues with Custody Image Retention

The Police National Database (PND), the main source for watchlist images, continues to hold millions of custody photos unlawfully. A 2012 High Court decision declared the retention of images from unconvicted individuals illegal, yet enforcement remains inconsistent.

In 2022, the National Police Chiefs Council (NPCC) estimated that approximately 18 million images might be retained in violation of legal standards, posing risks to police legitimacy and complicating the ethical use of facial recognition technologies.

Responding to these concerns, the NPCC initiated a nationwide review and management program for custody images in late 2023. However, as noted by the Biometric Commissioner Tony Eastaugh in 2024, many forces still use images of arrested but uncharged individuals for facial recognition, underscoring the need for clearer policies and stricter compliance.

Conclusion: Navigating the Complexities of LFR in Policing

While live facial recognition offers potential benefits for crime prevention, its current deployment in UK policing raises significant legal, ethical, and social challenges. The technology’s influence on police suspicion, the problematic nature of watchlist creation, and the unlawful retention of custody images highlight the urgent need for transparent governance, rigorous oversight, and respect for individual rights.

As LFR becomes more embedded in law enforcement, balancing technological innovation with civil liberties will be crucial to maintaining public trust and ensuring justice.