Sainsbury's pauses face-scanning AI after it wrongly accused a shopper of theft

Key points

• Sainsbury's paused facial-recognition AI at one of its UK stores after a false shoplifter identification in 2024.
• Customer Matt Arnold, 46, was ejected from the store and described feeling "embarrassed, mortified" and "powerless".
• The supermarket uses a system called Facewatch, which matches shoppers' faces against a watchlist of known offenders.
• Sainsbury's blamed "human error" rather than the technology itself for the wrongful ejection.
• The incident raises fresh questions about how AI identification tools should be used in everyday retail settings.

Matt Arnold walked into a Sainsbury's supermarket expecting to buy groceries. He walked out having been accused of shoplifting, a crime he did not commit.

The store's face-scanning system, a product called Facewatch, had flagged him as a known offender. Facewatch works by capturing images of every person who enters a store and comparing those images in real time against a private watchlist of people previously caught or suspected of theft. When the software finds a match, it alerts staff.

In Arnold's case, the match was wrong.

"I was embarrassed, mortified even, and felt quite humiliated and powerless," the 46-year-old told The Guardian, which first reported the story.

What is Facewatch, and how does it end up in a supermarket?

Facewatch is a facial-recognition service sold to retailers. Think of it as a digital wanted board: shops upload photos of people they have previously caught stealing, and the software silently checks every new face at the door against that list.

Sainsbury's is one of several UK retailers that have trialled the technology, hoping to cut shoplifting losses that the industry says have grown sharply in recent years.

The system works quickly and largely invisibly. Most shoppers have no idea they are being scanned.

So whose fault was it?

Sainsbury's says the technology is not to blame. The supermarket pointed to "human error" as the cause, suggesting a person made a mistake somewhere in the process, whether when adding someone to the watchlist or when acting on the alert.

That distinction matters, but it does not fully settle the concern. An AI identification tool that flags the wrong person still causes real harm, regardless of who pressed the final button.

Sainsbury's has paused use of the system at the store involved while it looks into what went wrong.

What does this mean for shoppers?

If you shop at a retailer using face-scanning technology, you may be scanned without being told, and without any clear way to challenge a false match before staff act on it.

Arnold's experience shows how quickly that can turn humiliating. There was no obvious appeal process at the door, no immediate explanation, and no way for him to prove on the spot that the system had made a mistake.

UK data protection rules require retailers to be transparent about facial recognition and to have a lawful reason to use it. The Information Commissioner's Office, the UK's data-privacy watchdog, has previously warned that the bar for using this kind of technology in public spaces is high.

For now, Sainsbury's says it remains committed to tackling retail crime, but the pause gives it time to review whether its safeguards are strong enough to protect innocent customers from the same ordeal Arnold went through.