Sainsbury’s has paused the use of AI-powered facial recognition cameras after a shopper was wrongly removed from one of its stores as a suspected shoplifter.

The company has suspended the technology involved while it investigates the incident, which took place at its Dulwich store in south London.

Matt Arnold, who works with the nearby Dulwich Hamlet Comedy Club, was waiting at a self-service checkout with a trolley of drinks and snacks when security staff approached him and told him the retailer could not serve him because of an alleged incident earlier in the week.

Arnold was subsequently asked to leave the supermarket and directed to contact Facewatch, the facial recognition technology provider used by Sainsbury’s.

The company contacted him the following day to confirm he had been wrongly removed from the store, attributing the incident to human error rather than a failure of the technology. Sainsbury’s apologised and gave Arnold a £150 voucher as a goodwill gesture, which he later donated to a food bank.

Arnold described the incident as a “terrifying glimpse of the future” and questioned the safeguards around the use of the technology.

It marks the second time this year that an innocent Sainsbury’s shopper has been wrongly removed from a store following an alert from its facial recognition system.

In February, Warren Rajah was escorted from the grocer’s Elephant and Castle branch after staff mistook him for an individual flagged by Facewatch. The technology provider later confirmed there were no alerts or incidents associated with Rajah, with Sainsbury’s again blaming human error and apologising.

Sainsbury’s recently unveiled plans to significantly expand its use of facial recognition technology as part of its efforts to tackle soaring levels of shoplifting and abuse against store workers.

It said last month it would install Facewatch cameras in up to 150 additional stores before Christmas, taking the roll-out beyond the 55 branches where it was already operating.

Sainsbury’s previously said trials of the system had resulted in a 46 per cent reduction in theft, harm, aggression and antisocial behaviour, while 92 per cent of identified offenders did not return to its stores.

Facewatch works by scanning shoppers entering participating stores and comparing their faces against a watchlist of people previously linked to theft, violence or aggression.

Store colleagues are alerted to potential matches, which are then supposed to be reviewed by a trained member of staff before action is taken.

The technology has faced mounting criticism from privacy and civil liberties groups, which have raised concerns over false identifications and the growing use of biometric surveillance across the retail sector.

Sign up here to get the latest grocery and food news each morning