How a Canadian Grocery Chain Got Caught Using Facial Analytics on Shoppers

A suburban retail store’s automated doors slid open with a gentle mechanical sigh on a rainy British Columbian afternoon. Customers didn’t realize that something else, something quieter and invisible, was also greeting them as they wheeled carts past racks of bulk paper towels and discounted patio furniture. Not a cashier. Not a welcome. As people entered, a biometric system scanned their faces.

Canadian Tire, or more specifically, a few of its associate-owned stores, was the retailer at the heart of the dispute. According to a British Columbia Office of the Information and Privacy Commissioner investigation, FaceFirst’s facial recognition technology was used to identify alleged shoplifters. Visitors’ still photos were taken by cameras, which then transformed them into biometric identifiers and compared them to a watchlist.

The system’s stated purpose was to discourage repeat offenders. In reality, it silently captured thousands of regular consumers.

Retailers everywhere seem to be feeling surrounded. Some regions of North America have seen a rise in shoplifting, and employees are reporting more instances of harassment. Following reports of a decrease in theft, Sainsbury’s recently extended its facial recognition pilots in the UK. Similar systems are used at airports. stadiums as well. Apparently, supermarkets came next.

Grocery stores, however, are not the same. They are private areas. Toddlers and parents bargain over cereal selections. Seniors at the register counting coupons. Teens hanging around the freezer section. It’s difficult to ignore how routine the scene is as you watch it play out—fluorescent lights humming overhead, apples carefully arranged in pyramids—and how remarkable the surveillance is.

In 2023, the British Columbia inquiry found that customers had not provided meaningful consent. Many people were unaware that their biometric information was being processed. The business maintained that the technology was only intended for known criminals. However, the cameras collected faces without discrimination. Before throwing away the majority of the biometric templates they created, they recorded every person entering the store.

That technical detail is important. Data is processed even if it is swiftly erased. Furthermore, biometric information differs from a password. Your face cannot be reset.

Advocates for privacy cite examples from other places as warnings. Rite Aid acknowledged that its system in the US produced thousands of false positives, misclassifying consumers as shoplifters. A man was ejected from a supermarket in London after employees mistook a facial recognition alert. Even though the margin of error is statistically small, statistics seem meaningless when you’re the one being approached in aisle seven.

Retail executives frequently maintain that these systems are 99 percent accurate or higher. It’s possible that they are in controlled environments. However, supermarkets are not research facilities. The lighting changes. Cameras tilt. People wear hats, grow beards, and age. Diverse faces may be difficult for algorithms trained on small datasets to handle, which raises unsettling concerns about bias.

Many observers in the Canadian case were taken aback not only by the technology’s use but also by its covert application. There was insufficient or no signage. Only after the privacy commissioner’s report was made public did customers find out about it. Finding out that you have been scanned without realizing it is a minor but persistent infraction.

Retailers contend that they are reacting to actual dangers. The figures aren’t made up. Theft and abuse are reported at high rates by employee unions. Employees who experience violence at checkout counters are sympathetic. Security is important. How far stores should pursue it is the question.

On the other hand, investors appear realistic. Preventing losses increases margins. Efficiency is promised by technology. Whether widespread backlash would have a significant impact on consumer behavior is still unknown. When asked, the majority of shoppers shrug. Some people say, “I’m not shoplifting.” However, that reasoning that presumes innocence shields you from mistakes. Recent examples from around the world indicate otherwise.

Corrective actions, such as more transparent notice and more stringent adherence to provincial law, were mandated by the privacy commissioner’s findings. However, the wider cultural change is still present. Once limited to police databases and border crossings, facial analytics is beginning to permeate daily business.

We have the impression that we are pushing limits in real time. Biometric systems are being tested by grocery chains. Lawmakers are rushing to amend privacy laws that were drafted before machine learning became a mature field. Perhaps unwittingly, consumers are getting used to being examined when purchasing milk.

These days, when you walk into a store and see the little black domes placed close to the entrance, you can’t help but wonder what they see. A possible burglar? A candidate for a loyalty program? Or just data—processed, anonymized, momentarily stored, or erased?

These systems are neutral, effective tools, according to tech companies. However, behavior is shaped by tools. They change how employees engage with patrons and how patrons feel in public areas. Historically, a supermarket has been a place of everyday anonymity. You look around. You make the payment. You go.

Perhaps that quiet anonymity is eroding.

Given that some U.S. states are considering banning, it’s possible that Canada will follow suit and enact stricter regulations. Or maybe stricter regulations and more obvious signage will normalize the practice. The matter remains unresolved.

It is evident that more than just the bar code on a milk carton is now being scanned. Many customers discovered this fact long after their faces had already been processed by the system, in the pages of a privacy report rather than at the checkout counter.