Sainsbury’s pauses AI facial recognition at London store after customer wrongly accused of shoplifting

Sainsbury’s has temporarily suspended the use of AI-powered facial recognition technology at one of its London stores after a customer was wrongly identified as a suspected shoplifter.

The incident happened at the supermarket’s East Dulwich store, where 46-year-old Matt Arnold was shopping for supplies for a comedy event.

Arnold said he was approached by two store managers while using a self-service checkout and was told he could not be served because of an alleged previous incident.

He was asked to leave the store and said staff attempted to escort him outside.

Arnold said the experience left him feeling embarrassed, humiliated and powerless.

After leaving the store, he noticed an overhead CCTV screen showing a red circle around his face, which he understood to be an alert generated by the facial recognition system.

His shopping was eventually left at the store for a friend to collect and pay for.

Sainsbury’s head office later apologised to Arnold and temporarily paused the use of its AI-assisted Facewatchtechnology at the store while an investigation takes place.

The supermarket said the incident was caused by human error rather than the facial recognition technology itself.

Sainsbury’s said the Facewatch system has an advertised accuracy rate of 99.98%, with every match reviewed by a trained manager.

Facewatch also said its technology was not responsible for the incident, saying a correct alert had been sent to the retailer but was subsequently mishandled by staff in the store.

The company said the temporary suspension was a precaution while employees receive additional training.

The incident has nevertheless renewed questions about the use of facial recognition technology in shops and whether staff can become too reliant on automated systems.

Arnold has called for the technology to be suspended across all stores until its reliability and use by staff can be properly assessed.

He said the incident raised wider concerns about what could happen if someone who is more vulnerable or less confident was wrongly identified and challenged.

Sainsbury’s has been expanding its use of facial recognition technology as part of efforts to improve staff safety and tackle repeat serious offenders. The company previously said trials of the technology had reduced logged incidents and that it was extending the system to additional London stores.

The latest incident highlights the difficult balance between using artificial intelligence to improve security and ensuring that innocent customers are not wrongly treated as criminals.

While AI systems can help identify potential offenders, critics argue that an automated alert should never be treated as proof of wrongdoing and that trained staff must carefully verify any identification before taking action against a customer.