For Angela Lipps, a 50-year-old grandmother from Tennessee, the last year has been a nightmare of bureaucratic error. Her ordeal began when facial recognition software used by Fargo, North Dakota police incorrectly identified her as a suspect in a bank fraud case. The result was a six-month jail stay for a crime that occurred over 1,200 miles away.

Lipps was arrested at gunpoint in her home last July while babysitting her grandchildren. Extradited to North Dakota, she faced multiple felony charges. The case against her rested on a digital match; a detective noted her facial features, body type, and hairstyle aligned with a woman seen on surveillance footage using a fake military ID. Lipps maintained she had never set foot in the state.

Her attorney, Jay Greenwood, secured her release on Christmas Eve after providing bank records placing her firmly in Tennessee during the frauds. The Fargo police had not contacted her prior to the arrest, nor did they arrange or pay for her return home after the charges were dropped. A local nonprofit ultimately helped her get back to Tennessee, where she found she had lost her home, car, and dog while incarcerated.

This incident is a stark example of the human cost when investigative tools are treated as infallible. Similar errors have surfaced globally, from a UK man wrongly arrested for a burglary after a facial recognition mismatch to a Baltimore student searched by police after an AI system misidentified a bag of chips as a gun. For Lipps, rebuilding her life continues, with no apology from the department that jailed her.

Source: The Guardian