Artificial intelligence is moving beyond chatbots and search engines and into public spaces, with police in Australia testing technology that can identify people using their faces.

Western Australia has begun trialling live facial recognition technology that scans faces in public places and compares them with a police watchlist in real time.

So, how does it work?

Cameras capture people’s faces as they move through an area. AI software analyses facial features and creates a digital representation, which is then compared with images stored on a police watchlist.

If the system detects a possible match, it alerts police officers, who can then investigate and confirm the person’s identity.

The technology is intended to help police locate people wanted in connection with serious offences, find missing people and identify individuals who may pose a risk to the public.

During the first week of the Western Australian trial, police said the system scanned more than 130,000 faces, generating 33 alerts and contributing to 18 arrests. The watchlist contained about 4,000 people.

But the technology is raising concerns over what happens when AI gets it wrong.

Facial recognition can produce false positives, where an innocent person is incorrectly identified, as well as false negatives, where the system fails to recognise someone it is looking for.

Accuracy can also be affected by factors such as lighting, camera angles, image quality and age. Researchers have previously raised concerns about differences in performance across demographic groups.

What happens to everyone else?

This is where the privacy debate begins.

Unlike conventional CCTV, live facial recognition does not simply record what is happening. It can actively analyse the faces of people passing through a public space and compare them against a database.

Police say people who are not on the watchlist can have their images deleted quickly. Privacy advocates, however, argue that stronger safeguards are needed around who can be placed on watchlists, how the technology is used and how long information is retained.

The debate is becoming increasingly important as police forces explore AI for surveillance.

The technology may make policing faster and more efficient. But the bigger question is whether it can be used accurately, transparently and responsibly without turning everyday public spaces into places where everyone is automatically monitored.

AI may be able to recognise a face in seconds.

The real question is: should it?