Wrong photo, six months in jail: a facial-recognition mismatch in a bank-theft case
A Tennessee grandmother was arrested and jailed for months after an AI face-match tool tied her to a bank fraud suspect she didn't resemble — now she's suing for $10 million.
What happened: Angela Lipps, a 50-year-old grandmother in Tennessee, was arrested in July 2025 while babysitting a neighbor's kids and accused of a string of bank thefts in North Dakota she had never been near. Investigators had used AI facial recognition to match a fake ID photo from a bank-fraud suspect to one of Lipps' social media pictures. She was held for months in a Tennessee jail, then extradited to North Dakota on a $100,000 bond she couldn't pay. Charges were dropped on Christmas Eve after her lawyer produced bank records proving she was in Tennessee the whole time. She's now suing the city of Fargo and the detective who sought her arrest warrant for $10 million.
Why it matters: Banks increasingly lean on AI tools, including facial recognition, to chase down fraud and theft suspects from surveillance footage. This case shows what happens when that identification becomes the main evidence for an arrest instead of one lead among several. Lipps didn't match the suspect's build, facial features, or tattoos, according to her complaint, yet the mismatch wasn't caught before a warrant was issued. For an industry built on verifying identity, a tool that misidentifies people this badly undercuts the trust the whole system depends on.
How it works, plainly: West Fargo police ran the fake ID photo used by the real suspect through facial recognition software, which surfaced Lipps' photo as a possible match. Fargo detectives investigating related bank cases then wrongly assumed the actual bank surveillance photos, not just the fake ID image, had also been checked against Lipps. That assumption skipped a step: no one independently confirmed Lipps resembled the person seen on the bank footage itself before a detective sought her arrest warrant, according to the complaint and the department's own account.
The rollout: Fargo's then-police chief said the department found 'a couple of errors,' acknowledged it doesn't own its own facial recognition system, and has since banned the kind of cross-department tool use that led to Lipps' arrest, routing such requests through a state intelligence center instead. The lawsuit, filed after the case drew local and national attention, argues the department never properly trained officers on the technology's limits. It's still pending; the city says it won't comment on active litigation.