Spotlight on Facial Recognition Tech as NZ Shopper Wrongly Identified as Thief

A display for facial recognition and artificial intelligence is seen on monitors at Huawei's Bantian campus on April 26, 2019 in Shenzhen, China. Kevin Frayer/Getty Images

Facial recognition software, powered by artificial intelligence and used by retailers to combat shoplifters, has incorrectly identified a New Zealand woman as a previously trespassed shoplifter, raising questions over the technology’s accuracy.

Rotorua woman Te Ani Soloman said she is now “paranoid about being labelled as a thief” after being stopped while shopping at a Foodstuffs-owned New World supermarket in the city on her birthday on April 2.

After being erroneously identified by the software as a shoplifter who had been trespassed from the store, Ms. Soloman was challenged by staff and she was forced to remonstrate with them.

She claimed that even after showing identification to prove she wasn’t the alleged person, they didn’t believe her.

“It didn’t seem to change their mind which was already made up based on what they saw,” Ms. Soloman said.

Foodstuffs has been trialling the technology in 29 of its stores since the start of February and plans to continue for six months. The company said the trial came after 4,719 incidences of retail crime were recorded in its stores nationwide in the last quarter of 2023—which was up by 52 percent on the previous quarter.

“Shockingly, one of our security team was stabbed recently and our people are being punched, kicked, bitten, and spat at. We’re seeing over 14 serious incidents a week, including an average of two assaults,” Foodstuffs North Island chief executive Chris Quin told RNZ.

“All too often it’s the same people, coming back to our stores despite having already been trespassed, committing more crime, and often putting our team members and customers at risk of abuse and violence.”

In a February release on the trial, Foodstuffs said an independent evaluator had overseen the implementation of the technology and it had “briefed” the privacy commissioner.

Visitors walk past a stand with AI (Artificial intelligence) security cameras using facial recognition technology at the 14th China International Exhibition on Public Safety and Security at the China International Exhibition Center in Beijing . Nicolas Asfouri/AFP via Getty Images

Criticism of the Software Worldwide

Facial recognition technology has seen increasing use in various fields, not only in retail but in law enforcement and border security. However, it has been associated with several high-profile failures and controversies, which include racial bias.

It has also raised significant privacy concerns. In 2019, it was reported that the U.S. Immigration and Customs Enforcement (ICE) used facial recognition technology to scan millions of Americans’ driver’s license photos without their knowledge or consent.

Research has shown that these systems often perform less accurately on darker-skinned individuals, leading to concerns about inaccuracy and discrimination.

A 2022 study in the UK by the University of Cambridge’s Minderoo Centre for Technology and Democracy called for a ban on police using the software after the misidentification of mainly dark-skinned people.

“Historically, surveillance systems are used to monitor marginalised groups, and recent studies suggest the technology itself contains inherent bias that disproportionately misidentifies women, people of colour, and people with disabilities,” read the study’s report.

Serious Concerns: Privacy Commissioner

In New Zealand, the Privacy Commission has also raised concerns citing overseas failings.

Commissioner Michael Webster told RNZ the technology is “not a proven tool,” and he is worried brown-skinned people could be unfairly targeted.

“I don’t want to see people incorrectly banned from their local supermarket and falsely accused,” Mr. Webster said.

“New Zealanders deserve to shop for their milk and bread without having their faces scanned unless it’s really justified.”

“We wouldn’t accept being fingerprinted and checked at the door before shopping for groceries—that sounds ludicrous—but FRT is a similar biometric process that is faster, machine-run, happens in a nanosecond and creates a template to compare your face to, now and in the future,” he added.

Appearing on the TVNZ Breakfast show, Ms. Soloman was asked whether she thought her race was to blame for the error.

In response, she said it was a “huge factor, ”adding that after the incident, she didn’t think the software should be used by supermarkets.

“Unfortunately, it will be the experience of many Kiwis if we don’t have some rules and regulations around this [technology],” she said.

“It’s made me, unfortunately, explain to my son what racism is.”

She has received an apology from Foodstuffs, which she described as “stock standard.”

“[It’s] ironic they blame human error for an AI piece of technology knowing it will have false positives and errors across the board, ” Ms. Soloman said.

In a statement to the TVNZ, a Foodstuffs spokesperson said Ms. Soloman “has every right” to be upset.

“We can completely understand her reaction. When we make a mistake we own it and act quickly to put things right, and this morning the store has undergone an operational audit and all appropriate team members will receive further training on required protocols.

“We apologise to the customer and will be contacting her directly to apologise for what has happened.”