WA police end facial-recognition trial after 1.38 million scans and 122 arrests

Western Australia Police have withdrawn their facial-recognition vans from public deployment after a trial that police say scanned 1,380,882 faces and led to 122 arrests. The five-month exercise also generated nine incorrect alerts, prompting questions about how police describe accuracy as the force begins an internal review of performance, safeguards and community impacts.

A watchlist check across public places

The trial covered 100 deployments at 43 metropolitan and regional locations. Cameras mounted on or near marked police vehicles scanned faces in designated public areas, comparing them with an alert list drawn from police databases. That made the system a live check of people passing through those areas, rather than a search of photographs after an incident.

The Perth Royal Show illustrates the scale of an individual deployment setting. Across three operations there, police said cameras scanned 223,749 faces, generating 15 alerts. Those operations resulted in three arrests and seven interactions with registered sex offenders.

Arrests were not the only reported outcome. Across the full trial, police also said alerts led officers to:

• Engage with 146 registered sex offenders.
• Locate 10 people for welfare checks.

An earlier official update, first published September 22, recorded 905,117 scans, 79 arrests and eight incorrect alerts. It said deployments began June 22. Those interim figures provide a progress snapshot; the newly reported totals cover the completed trial.

Nine false alerts, two different yardsticks

Police expressed the nine incorrect alerts as a share of all scans, producing a rate of 0.00065%. That figure describes false alerts relative to scanning volume. It does not describe the share of alerts that were wrong: its denominator is every face scanned, not every alert generated.

Commissioner Col Blanch acknowledged that the technology was not perfect and said police knew that before the trial. He defended following up an alert as comparable to responding when a member of the public raises suspicions about someone. Officers spoke with each of the nine people incorrectly flagged.

Civil Liberties Australia chief executive Bill Rowlings challenged the framing, accusing police of “severely gilding the lily”. He argued that the arrests should be judged against eventual convictions, saying the public would only know the incorrect rate after those arrested had their day in court. His proposed measure concerns court outcomes, rather than the false-match calculation police presented.

No one was arrested. No one made a complaint. Each one did result in an interaction with that person and asking them relevant questions.

Col Blanch, Western Australia police commissioner, on the false alerts

Deletion safeguards and collective privacy

Police said images of people not on the alert list were pixelated in real time and immediately deleted without being stored. The official progress update also described that safeguard, saying there was no record-keeping of those images. It was the force’s stated protection for people scanned who were not targets of an alert.

The privacy objection went beyond whether images were retained. Rowlings said the scans were conducted with no reasonable suspicion. When the trial was announced, Western Australia’s independent privacy watchdog, the Office of the Information Commissioner, warned of “collective privacy harms” and said it had not been involved in implementing the exercise.

The force’s internal evaluation now includes how the technology performed on the ground and its community impacts. Police said that work would include reviewing the safeguards and whether all requirements were met. The vans remain out of public deployment while that assessment takes place.