Western Australia Police have launched Australia’s first public-facing trial of real-time AI facial recognition, using cameras in Perth and Fremantle to scan pedestrians’ faces in real time and compare them against police watchlists, raising concerns over privacy and the expansion of surveillance powers.
Police began using technology company NEC’s NeoFace m40 system in June. In its first week, the system scanned more than 130,000 faces and compared them against a watchlist of around 4,000 people. The list includes people suspected of serious crimes, missing persons, and individuals police believe may pose a risk to themselves or others.
Police said the trial had delivered encouraging initial results, triggering 33 alerts and contributing to 18 arrests. Authorities said real-time facial recognition could help officers identify people of interest more quickly, improving public safety and law enforcement efficiency.
However, privacy experts and legal advocacy groups have questioned how police decide who is placed on the watchlist, how the data is stored, and the risk of misidentification, arguing that there is insufficient transparency. The choice of trial locations has also attracted scrutiny, with concerns that the technology could result in disproportionate surveillance of Indigenous communities.
Unlike “one-to-one” facial recognition, which is used simply to verify an individual’s identity, real-time “one-to-many” systems can scan and compare faces on a large scale in public places.