New Data On London Metro Police Facial Recognition Tech Shows It's Still Wrong 96 Percent Of The Time

The Metropolitan Police’s facial recognition technology has demonstrated a catastrophic failure rate, consistently misidentifying innocent citizens as criminals. Rather than confirming accuracy, Freedom of Information requests reveal that the vast majority of matches are false positives, with some instances showing a near-total inability to distinguish between the public and wanted individuals. This persistent inaccuracy suggests the system is fundamentally unreliable for law enforcement purposes, casting serious doubt on its practical utility. The human cost of these technical failures is significant, leading to wrongful stops and investigations of innocent people, including minors. Despite evidence of systemic error, police continue to deploy the technology, accumulating non-criminal biometric data in the process. While retention policies have been slightly adjusted to delete non-matching images sooner, the core issue remains unresolved: the system effectively criminalizes ordinary public interaction through algorithmic error, raising serious civil liberty concerns. This situation is critically relevant to open data discussions as it highlights the dangers of deploying opaque, error-prone algorithmic systems without rigorous transparency or accountability. The reliance on flawed automated decision-making underscores the need for strict governance around biometric data collection and usage. It serves as a cautionary tale for how open or accessible data frameworks can inadvertently facilitate surveillance overreach when technological safeguards and ethical oversight are insufficient, demanding greater scrutiny of how public institutions utilize automated identification tools.

Source: techdirt.com
Published on 2024-02-15

New Data On London Metro Police Facial Recognition Tech Shows It's Still Wrong 96 Percent Of The Time

Related news