Call for Urgent Oversight as WA Police Live Facial Recognition Trial Scans 130,000 Citizens

A ground-breaking trial of real-time artificial intelligence facial recognition by a police force has drawn sharp criticism from legal advocates and privacy experts. The mass surveillance pilot, currently underway in Western Australia, is being closely watched by law enforcement agencies nationwide.

Since launching in June, the trial has used cameras linked to advanced facial recognition software to scan more than 130,000 individuals across Perth and Fremantle. These scans compare members of the public against a police watchlist in real time. While local authorities describe early outcomes as promising, the program has triggered serious concerns regarding an opaque watchlist and locations that critics argue may disproportionately target First Nations populations.

How the Technology Works

The system operates by scanning faces in public spaces and comparing them to a police database using equipment housed in a nearby van. This represents a significant shift into complex "one-to-many" systems, which compare individual faces against a larger database of flagged individuals. This is far more advanced than "one-to-one" systems, such as those used to unlock personal mobile phones, or private sector surveillance used to flag suspected thieves or abusive customers.

Utilising technology developed by NEC, specifically the Neoface m40 product, the pilot scanned over 130,000 faces in its first week alone. These images were matched against a watchlist of approximately 4,000 individuals, including those accused of serious crimes, missing persons, or individuals deemed a risk to themselves or others. Authorities celebrated the initial results, which generated 33 alerts and led to 18 arrests. Officials state that images of individuals who do not match the database are pixelated and deleted almost immediately.

However, during that first week, the algorithm falsely identified one individual—a rate that officials note outperformed a similar 2023 trial of the technology in the UK. While the pilot has continued for several weeks across various locations, police have declined to share subsequent data, indicating that further details will be released in a coordinated manner.

Demographic Disparities and Technical Limitations

Experts have warned that facial recognition systems carry inherent accuracy flaws. Associate Professor Jake Goldenfein from Melbourne Law School pointed out that these algorithms historically perform poorly on specific demographics. Historically, these systems have performed worse for people of colour and for women, and especially worse for women of colour, he said. This discrepancy occurs because the algorithms are trained predominantly on datasets of white, male faces.

According to Dr. Goldenfein, these systems produce both false positives and false negatives. While police pointed to a 2023 UK National Physical Laboratory study showing no statistically significant differences in false positives by gender or ethnicity under specific conditions, that same study found that accuracy degraded when system settings were relaxed or when real-world factors—such as lighting, picture quality, skin tone, and age—were introduced.

Michael Birtwisle, an associate director at the London-based Ada Lovelace Institute, cautioned against relying solely on British data, especially given the lack of testing on First Nations populations. It's a snapshot so it can only tell you [about] that version of that system, for that population, Mr. Birtwisle said.

Local authorities did not clarify whether the NEC algorithm had undergone testing on First Nations faces before the trial. However, a police spokesperson stated: WA Police is committed to transparency and is collating statistics throughout the trial, adding that The trial will assess demographics during the deployments and gauge whether or not there is a disproportionate impact on an ethnicity or group within the community.

Legal Advocates Left in the Dark

The Aboriginal Legal Service of Western Australia (ALSWA) has criticised the lack of consultation surrounding the rollout. Peter Collins, the organisation’s legal services director, revealed he received an email from a police superintendent while on leave, requesting just 20 minutes to discuss the imminent trial. This is the extent of the consultation: an email asking to meet two days before it is launched, Mr. Collins said.

Mr. Collins argued that the state's primary legal body for First Nations people should have been deeply involved, particularly given high incarceration rates and the risk of over-policing vulnerable groups. Instead, he described the interaction as an exercise in tokenism. He added: This was a done deal from the get go. No matter what ALSWA had to say about the pilot it would have gone ahead, noting that authorities have also failed to clarify if the technology scans children. Police spokespeople countered that they had made attempts to meet with ALSWA.

Growing Concerns Over Surveillance

In its preparatory privacy impact assessment, police declared the trial would not primarily or especially affect Aboriginal people. However, Mr. Collins disputed this, arguing that the pilot represents an escalation in the over-surveillance of local Aboriginal communities. It can only have been by design, not chance, that facial recognition technology has been rolled out in those suburbs and places where everyone is aware that significant numbers of Aboriginal people either live or frequent, he said, adding, It is easy to understand why many Aboriginal people in WA believe they live in a police state.

A police spokesperson defended the choice of locations, stating that the locations are identified ... based on policing need and policing objectives. Locations do not target any group of people or demographic.

John Pane, a privacy expert at Electronic Frontiers, expressed concern over the legislative framework surrounding the technology. It grants extraordinary powers, with limited oversight, to a police agency to do incredibly invasive things, Mr. Pane said, calling for urgent updates to privacy legislation.

Additionally, Western Australia's privacy watchdog, the Office of the Information Commissioner, stated it was not invited to engage in any meaningful consultation regarding the trial, despite police claiming they briefed the office in April.