Challenges and Considerations in Live Facial Recognition Trials by European Law Enforcement

Recent research from the University of Cambridge has raised significant concerns about the deployment of live facial recognition (LFR) technology by police forces across the UK and Europe. The study describes current real-world testing as a “Wild West” scenario, where these systems are trialed on local populations without sufficient oversight, safeguards, or ethical governance.

Reevaluating the Nature of Live Facial Recognition Trials

While field testing of AI-driven systems like LFR offers valuable insights into their operational capabilities, existing trials have largely overlooked the broader socio-technical implications. A comparative analysis of LFR deployments in London, Wales, and Berlin reveals that these tests often fail to provide robust evidence of tangible benefits, such as improved crime detection or public safety outcomes.

Experts Karen Yeung, a professor specializing in law, ethics, and informatics, and Wenlong Li, a research professor in law, emphasize the urgent need for clear regulatory frameworks and governance to ensure that LFR trials are conducted responsibly-both legally and ethically. Without such structures, these trials risk becoming mere public spectacles designed to legitimize invasive surveillance technologies without meaningful public discourse or accountability.

Privacy, Rights, and the High Stakes of Facial Recognition

Yeung and Li stress the importance of acknowledging the profound privacy risks and potential for misuse inherent in LFR technology. Given its capacity for mass surveillance and intrusion, any deployment must meet exceptionally stringent standards to justify its impact on fundamental rights such as privacy, freedom of expression, and assembly.

Without comprehensive and transparent evaluations of LFR’s societal effects, there is a danger that incremental erosions of civil liberties could occur unnoticed, undermining the freedoms essential to democratic societies. Vigilance is crucial to ensure that technological advancements do not come at the expense of lawful personal autonomy and social development.

Questioning the Legitimacy of Police LFR Trials

Examining the Metropolitan Police Service’s (Met) LFR deployments from 2016 to 2020, Yeung and Li argue that labeling these operations as “trials” is misleading. These deployments closely resembled actual policing activities with real legal consequences for individuals flagged by the system, rather than controlled experiments designed to assess the technology’s broader impacts.

These so-called trials primarily focused on the system’s ability to generate alerts leading to arrests, without adequately evaluating the socio-technical processes or the human factors involved. For instance, the Met has yet to fully integrate the necessary organizational and human elements required to translate automated facial matches into lawful arrests, which legally require reasonable suspicion-a threshold that an LFR alert alone does not satisfy.

In England and Wales, police can stop and question individuals but cannot compel answers without reasonable suspicion of criminal involvement. Therefore, any engagement based solely on an LFR alert must respect this legal boundary.

The “Presumption to Intervene” and Its Implications

Despite legal safeguards, previous evaluations of London’s LFR trials revealed a troubling “presumption to intervene,” where officers felt compelled to act on algorithmic prompts regardless of contextual judgment. This highlights the necessity for clear operational policies, comprehensive officer training, and strict protocols to prevent overreach.

Moreover, while the Met claimed additional benefits such as crime deterrence and disruption, these assertions lack empirical support. Independent reports, including one from the National Physical Laboratory (NPL) in 2019, suggested potential safety improvements without providing concrete evidence. Similarly, South Wales Police (SWP) conducted 69 operational trials between 2017 and 2020, but independent researchers from Cardiff University were unable to quantify any impact on crime prevention.

Concerns were also raised about the quality and consistency of watchlists used in these trials, with significant variability in image quality and size, undermining the reliability of the system. Researchers recommended transparency regarding watchlist composition to enable more rigorous evaluation.

Importantly, these trials could only measure detected matches (true and false positives), leaving unknown the number of individuals who passed undetected (“false negatives”), a critical gap in assessing overall system effectiveness.

Need for Comprehensive Socio-Technical Assessments

Trials conducted in cities like Nice and Berlin involved volunteers who consented to participation, contrasting with the random public testing seen elsewhere. These studies primarily assessed technical performance rather than operational or societal impacts. For example, in Berlin, some passersby were unknowingly included in data collection, raising ethical concerns about informed consent and data privacy.

Despite more controlled conditions, these trials failed to demonstrate clear real-world benefits such as improved efficiency in suspect identification or apprehension. Yeung and Li conclude that law enforcement agencies have yet to fully grasp the complex ethical and legal challenges posed by LFR deployment.

They argue that for invasive biometric surveillance technologies, especially those capable of remote monitoring like live facial recognition, rigorous evidence of societal benefit is essential. This evidence must satisfy legal standards of necessity and proportionality to justify any infringement on fundamental rights.

Building public trust in LFR requires more than just technical accuracy; it demands thorough integration into organizational workflows and human decision-making processes, ensuring that the technology’s use aligns with social values and respects civil liberties.

Official Responses from the Met and South Wales Police

In response to these critiques, a spokesperson for the Metropolitan Police stated confidence in the proportionality and legality of their LFR use, acknowledging that early operational testing was limited. The Met has since engaged the National Physical Laboratory for independent assessments, which have informed more equitable deployment strategies.

Since early 2023, the Met has expanded LFR use in identified crime hotspots across London, reporting strong community and stakeholder support. According to recent figures from 2025, the system has scanned over 2.5 million faces, resulting in more than 1,300 arrests since 2024, with only 12 false alarms recorded.

Public opinion surveys commissioned by the Mayor’s Office for Police and Crime in late 2024 indicate that 84% of respondents back the use of facial recognition technology to identify violent offenders, serious criminals, and individuals wanted by courts or at risk themselves.

South Wales Police also emphasized that their LFR testing was independently evaluated by the National Physical Laboratory, underscoring their commitment to transparency and accountability.