Rise of AI: NZ supermarkets using facial recognition

Foodstuffs NZ uses facial recognition CCTV technology in some North Island stores. Photo: Getty...
Foodstuffs NZ uses facial recognition CCTV technology in some North Island stores. Photo: Getty Images
Inquiries about a Dunedin man mistakenly identified as a shoplifter at New World have led to the revelation that New Zealand’s largest supermarket company  has quietly rolled out facial recognition CCTV technology in some of its North Island stores.

The man was allegedly mistakenly identified due to human error, and Foodstuffs NZ claimed facial recognition was not used in the South Island. However, the Otago Daily Times can reveal a different security system that "bridges the gap between businesses and the police" is now used at the Centre City New World in Dunedin, among other South Island stores.

Dunedin mechanic Daniel Ryan said he was recently taken aside by staff shortly after entering the Centre City New World in Great King St, owned by Foodstuffs. He alleged he was taken into a side room and questioned by staff, who said he had been identified as a known shoplifter.

Mr Ryan said the staff then realised he had been mis-identified and he received an apology from the company. While he appreciated the apology, the experience left him feeling humiliated.

"It’s quite bruising to be shuffled off to the side."

Foodstuffs head of external relations Antoinette Laird said "human error" had led to Mr Ryan being mistakenly identified as a shoplifter. Asked if Centre City New World was using a facial recognition surveillance system, Ms Laird said the technology was used in some of its stores, but none in Dunedin.

"A handful of stores in the North Island have facial recognition CCTV technology as part of their security system.

"We cannot provide specific store detail."

Facial recognition technology is widely used by retailers overseas.

The Guardian has reported 59% of fashion retailers in the United Kingdom use facial tracking, which captured the faces of shoppers, before cross-referencing their biometric data with known criminals.The technology is also prevalent in China, where local governments use it to track people in public places.

Foodstuffs owns the New World, Pak’n Save and Four Square supermarkets.

While Centre City New World did not use automatic facial recognition, the store employed the"Auror" security system, Ms Laird said.

"This system captures images, licence plate numbers etc, enabling our loss prevention staff to identify offenders more easily and get on top of theft."

Started in New Zealand, Auror offers software platforms designed to"help police and retail businesses collaborate and fight crime," according to its website.

Auror content and communications manager Kevin Ptak said the company’s software used images from  CCTV cameras to track"repeat offenders".

"It’s used in the back office."

Z Energy was also an Auror client, employing its software to automatically identify the licence plates of vehicles linked to fuel theft drive-offs and alert attendants.

The company’s name was inspired by the Harry Potter series of books by J. K. Rowling, Mr Ptak said.

In the books, Auror is a title used by witches and wizards tasked with magical law enforcement.

The Office of the Privacy Commissioner was not aware of any supermarkets in New Zealand using facial recognition technology before the ODT informed it of the practice last week.

A spokesman for the commissioner said he strongly encouraged supermarkets considering using the technology to undertake a"privacy impact assessment," and urged anyone unhappy about  having their face automatically identified to speak up.

"We would expect to see signage and messages informing customers that the technology is in use, and what their information will be used for.

"If individuals feel their privacy has been breached by this technology, they should complain to the supermarket first. If they are unsatisfied with the outcome of that complaint, they can complain to our office."

george.block@odt.co.nz

Add a Comment