Grandmother sues after AI facial recognition error led to 6-month wrongful detention

Angela Lipps was arrested at gunpoint on a July afternoon in 2025, surrounded by U.S. Marshals at her neighbor's house in Tennessee. She was caring for four young children at the time. The warrant that brought them to that door was based on a single piece of evidence: a match generated by artificial intelligence facial recognition software, comparing her face to a photo pulled from a fraudulent ID card used in a bank theft case 800 miles away in North Dakota.

Lipps, now 50 and from Elizabethton, Tennessee, spent 108 days in a county jail without bail. No one interviewed her. No one reviewed her financial records, which would have shown she was in Tennessee when the bank thefts occurred in April and May 2025. She was charged with eight felony counts, the most serious carrying a potential sentence of ten years in prison. On December 24, 2025—six months after her arrest—prosecutors dismissed all charges after she produced the bank records proving her whereabouts.

The lawsuit Lipps filed on September 15, 2026, against the city of Fargo and its police department alleges a cascade of investigative failures that transformed an AI tool into an instrument of wrongful detention. The bank theft case itself involved an unknown woman using fake identities to commit fraud across Fargo and West Fargo. Rather than using surveillance footage from the banks themselves, Fargo police conducted their facial recognition search using Clearview AI and a photograph taken from one of the fraudulent ID cards. Clearview AI's own guidelines explicitly state that facial recognition matches should be treated as potential leads requiring independent corroboration—not as positive identifications.

Detective Lucas Heck, according to the complaint, characterized the AI match as a positive identification without disclosing that it had been uncorroborated. He did not conduct photo lineups with bank tellers. He did not review Lipps' travel records. Surveillance video from the banks showed physical differences between Lipps and the actual suspect: a different build, and a neck tattoo on Lipps that was absent from the woman in the footage. The West Fargo Police Department, reviewing the same AI match, declined to pursue charges precisely because it lacked independent corroboration. Heck's department proceeded anyway.

When Lipps was released, she discovered that her residence, her car, and her personal property had been lost during her six-month absence. The Fargo Police Department and city acknowledged investigative errors at a press conference in March 2026. Mayor Tim Mahoney stated that when errors in the facial recognition process were identified, the department addressed them immediately.

Lipps' case is not isolated. Clare Garvie, deputy director of Technology Law & Policy with the Policing Project, told Fox News Digital that at least 16 documented wrongful arrests have stemmed from facial recognition misidentifications where individuals spent time in jail. In many of those cases, police failed to take the investigative steps necessary to verify or disprove the match. Nate Freed-Wessler, a deputy director at the American Civil Liberties Union, said wrongful arrests from AI technology have been occurring for at least five or six years, and the rate appears to be increasing as law enforcement agencies integrate these tools into investigations without updating their protocols or oversight mechanisms.

Garvie noted that the challenge lies in how police work itself must evolve. Officers are being asked to do a different job now—to work with AI as an investigative tool rather than as a substitute for investigation. That shift requires rethinking workflows, training, and accountability at every stage. The question facing police departments across the country is whether they will implement those changes before the next person is arrested at gunpoint based on a match that should have been treated as a lead, not a conviction.