Facial Recognition: Public Safety Tool or Privacy Threat?

Western Australia has launched an Australian-first five-month trial of mobile, real-time facial recognition technology for law enforcement. Mounted on police vans, the system scans hundreds of faces per minute, cross-referencing them against a database of individuals with warrants, child sexual abuse offender registrations, and missing persons. During its initial week, a single van scanned over 130,000 faces, leading to 33 alerts, 19 arrests, two welfare checks, and engagement with 21 registered sex offenders.

Despite WA police hailing the technology as a cutting-edge public safety tool, it has faced criticism from the state's Office of the Information Commissioner (OIC WA) and cybersecurity experts. Concerns include potential algorithmic bias, accuracy issues in variable conditions, and 'function creep' – the risk of the technology's scope expanding beyond its intended use. The OIC WA stated it was not involved in the trial's implementation and warned of collective privacy harms, disputing claims that the system avoids mass surveillance.

Experts highlight that facial recognition accuracy can drop significantly in uncontrolled public environments, and bias in algorithms may disproportionately affect marginalized groups, including Aboriginal and Torres Strait Islander communities. The lack of dedicated federal or state legislation for live biometric surveillance exacerbates these concerns. The trial is being monitored by the OIC WA to ensure compliance with the new Privacy and Responsible Information Sharing Act 2024, Australia's first law regulating automated decision-making.