This article contains affiliate links. If you buy through them, we may earn a commission at no extra cost to you.
Angela Lipps was babysitting her neighbor’s four children in Tennessee when U.S. marshals showed up at her door with guns drawn. She had never committed a crime in North Dakota. She had never even been on a plane. She had no idea why they were there. Six months later — after losing her residence, her car, and chunks of her life she will never get back — she walked out of a North Dakota detention center on Christmas Eve into freezing cold with no way to get home. Now, according to ABC News, she’s suing for $10 million. The reason any of this happened: an AI facial recognition match that police used as the primary basis to arrest and extradite her across state lines for crimes she didn’t commit.
- Angela Lipps, 50, is a grandmother of five and mother of three from north-central Tennessee.
- She was arrested in July 2025 and detained for nearly six months before charges were dropped.
- Bank records proving she was in Tennessee during the alleged crimes were available — and produced by her attorney — but came too late to stop the arrest.
- The Fargo Police Department and detective Lucas Heck are named defendants in the federal lawsuit.
- Facial recognition misidentifications have resulted in at least eight documented wrongful arrests nationally, with research showing the technology fails most often on people of color, women, children, and the elderly.
A Tool That Doesn’t Work Is Being Used Like One That Does
Facial recognition technology has a documented accuracy problem. It is not subtle. It is not disputed. Researchers have shown for years that these systems misidentify people of color, women, and older individuals at far higher rates than white men. That is not a fringe finding — it is the scientific consensus. And yet law enforcement agencies across the country keep feeding it into arrest warrant applications like it’s a fingerprint match.
Fargo police, according to the complaint, knew that facial recognition results should not serve as the sole basis to extradite and charge someone. They did it anyway. A 50-year-old grandmother was arrested at gunpoint, booked into county jail, held for nearly four months without bail, flown across the country, and charged with eight counts. The exculpatory evidence — basic bank records — was sitting right there the entire time. Nobody went looking for it until Lipps and her attorney forced the issue.
This is not a story about AI failing. This is a story about humans choosing to trust a flawed tool over basic investigative work because it was faster and easier. The AI didn’t wrongfully arrest Angela Lipps. A detective signed a warrant. A city backed it. A system let it happen.
The New Jersey Supreme Court recently ordered prosecutors in a separate murder case to disclose exactly what facial recognition software was used and how investigators applied it — a ruling that builds on earlier precedent requiring transparency around the technology. That’s progress. But disclosure after the fact doesn’t un-arrest anyone. It doesn’t return a person’s home or car or six months of their life.
Why Does This Keep Happening?
Because there is almost no federal regulation governing how police can use facial recognition. Because departments adopt the technology faster than any oversight structure can follow. And because until lawsuits like Lipps’ start costing cities real money, there is minimal institutional pressure to stop.
We’re at a strange moment in the AI story. On one hand, tools like personal AI agents are being positioned as helpful, human-centered technology designed to empower everyday people. On the other, governments are deploying AI systems against those same everyday people with almost no accountability when those systems get it catastrophically wrong. The gap between those two realities is enormous — and Angela Lipps fell straight through it.
Companies like Surfshark, which recently moved into personal data removal, understand that people’s digital identities carry real stakes. A face is the most personal data a person has. Turning it into probable cause without safeguards isn’t just a civil liberties concern — it’s a live threat to anyone who happens to share a resemblance with a suspect and lives in the wrong zip code.
The $10 million figure will get headlines. But the actual ask buried in that lawsuit is simpler and more damning: make the Fargo Police Department answer for what it did to a grandmother on her neighbor’s couch, in broad daylight, with evidence of her innocence already in existence.
The real cost of cheap AI shortcuts isn’t measured in processing time — it’s measured in the six months Angela Lipps will never get back.
Watch the Breakdown
Sources
- Grandmother falsely accused of bank theft after alleged AI facial recognition match sues for $10M — abcnews.com
- US woman falsely charged with bank theft sues for $10m over AI facial recognition — www.theguardian.com
- Top court orders disclosures in NJ cops’ use of facial recognition technology — newjerseymonitor.com
