Sainsbury’s has temporarily suspended live facial recognition at its East Dulwich store after staff wrongly challenged a customer as a suspected shoplifter and asked him to leave.
Matt Arnold, a 46-year-old comedy promoter, was using a self-service checkout when two managers told him the store would not serve him because of an incident earlier in the week. After leaving, Arnold said he saw a security monitor displaying his face inside a red circle.
Sainsbury’s subsequently contacted Arnold to apologize. The retailer said the incident resulted from human error rather than a failure of the facial recognition system, which is supplied by Facewatch. It paused the technology at the branch while the incident is reviewed.
Facewatch offered a similar account, saying its system had generated a correct alert but that the retailer handled and communicated it incorrectly. Sainsbury’s said every potential match is reviewed by a trained manager before staff take action.
The incident occurred as Sainsbury’s expands a program that began with an eight-week facial recognition trial at stores in Sydenham and Bath. The system captures facial images at store entrances and compares them with a watchlist of people associated with previous incidents of theft or aggression.
Customers entering participating locations are informed through signs that facial recognition is in operation. Sainsbury’s says it will continue to provide refresher training to staff as it develops its use of the technology.
The East Dulwich case is not the first time an innocent shopper has been challenged at a Sainsbury’s store following a Facewatch alert. In an earlier incident, staff at the Old Kent Road branch escorted another customer from the premises after approaching the wrong person associated with an alert.
UK retailers have increasingly adopted biometric systems as part of efforts to address shoplifting and violence against workers. Iceland has also tested Facewatch at selected stores, while several other chains use the platform for loss prevention.
The suspension illustrates the role of human review in biometric alert workflows. Even when a system identifies the intended watchlist subject, staff must still verify that they are approaching the correct person and communicate the basis for any action accurately.
Sources: LBC
–
By Ali Nassar-Smith
