WA police is using live facial recognition to make arrests. This trial is testing privacy law
Who gets to authorise such deployments? And who checks the safeguards?
In Western Australia, police have become the first law enforcement agency to employ live facial recognition technology for identifying persons of interest. A clearly marked van equipped with cameras roams Perth and Mandurah, scanning the faces of everyone it passes and comparing them against a watchlist of around 4,000 individuals, including those with outstanding arrest warrants, reportable offenders, people subject to lawful exclusion orders, and missing persons.
The trial is open and publicly advertised, with deployment dates announced beforehand. During its first week, the technology scanned over 130,000 faces and generated 33 alerts, leading to 18 arrests and interactions with registered sex offenders. This process, which would have taken weeks using traditional investigative methods, was completed in just days by a single van.
However, the use of this technology raises several concerns regarding governance and privacy. The trial's effectiveness hinges on pre-defined success criteria, independent evaluation, and potential for non-permanent deployment. In WA, the police department both conducts the deployment and compiles the results, and no independent evaluator has been appointed.
The privacy impact assessment indicates that funding has not yet been determined, and event security is listed as one of the intended uses. While the arrests are seen as a measure of success, they are not crimes solved but rather a rapid review of a known watchlist. The system also results in false positives, with an error rate of 3-6%, significantly higher than the 0.0008% reported by WA police.
The bias in facial recognition algorithms, which tends to produce higher false-match rates for women, younger individuals, and people with darker skin, adds another layer of concern. In a policing context, this raises the risk of wrongful stops for certain groups. The system used, NeoFace Watch from NEC, has been benchmarked by the UK's National Physical Laboratory (NPL) for bias, but no results from WA's deployment have been independently verified.
The algorithm automatically deletes biometric data of non-matches and shares no data with third parties before any action is taken. However, there are no independent audits of these safeguards, and it's unclear if the claimed data deletion actually occurs.
Governance is crucial for the responsible use of such technology. In the UK, a court ruled South Wales Police's use of live facial recognition unlawful. Despite Australia lacking a national legal framework for biometric surveillance, WA's trial now serves as a national test case for such legislation. The state's Privacy and Responsible Information Sharing Act 2024 mandates that WA police demonstrate how risks of bias, harm, and discrimination are managed.
However, the absence of independent scrutiny and the lack of a national legal framework for governing biometric surveillance leave many unanswered questions about the compliance of WA police's use of live facial recognition with international human rights law.
Written by urgent.news from The Conversation AU's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
