Face Bank: Wrong Identity Match Jails Grandmother 6 Months

Angela Lipps was babysitting her neighbor's kids when the police showed up. She is 50, a grandmother of five, and she says a face bank search (a stored collection of face photos that software compares against) pointed straight at her. She spent six months locked up for a bank theft in a state she says she'd never been to.

Face bank matches are tips that need checking, and Angela Lipps' $10 million lawsuit shows the damage when a computer's guess gets treated like proof.

What Is a Face Bank, and How Did One Mistaken Identity Cost a Grandmother Six Months?

Let's start with the plain version. A face bank is a big pile of face photos. Software takes a new picture, compares it to the pile, and hands back the faces that look most alike. That's it. The software doesn't know who did what. It only knows which faces resemble each other.

According to ABC News, West Fargo police ran a photo from a fake ID through facial recognition software. It spat out a match to Lipps' social media photo. Police arrested her in July 2025. The suspect in the fake ID photo, by the account of her lawsuit, didn't resemble her in build, facial features, or tattoos.

Read that again. This was not a blurry-camera, close-call situation. A person looking at the two photos side by side, the way you'd compare two pictures on your kitchen table, apparently should have seen the problem.

Face Banks and the Digital Paper Trail That Actually Freed Her

Here is the part that should stick with you. The software did not free Lipps. Old-fashioned records did. Her bank account showed she was in Tennessee when the thefts happened. That digital paper trail (a record of where your card was used and when) was proof. The face match was only a guess.

By then she had lost her home and her car. So the $10 million she is asking for is not really about the software. It is the price tag on skipping one boring step: checking whether anything else supports the match.

Why a Face Bank Match Is Not Proof: The Account of Who Skipped the Checking

Why does this keep happening? Our brains are the problem as much as the machines. When you hear "AI matched her face," it sounds scientific. It sounds finished. Psychologists call this the availability heuristic: whatever story comes to mind fastest feels most true. "Computer says so" comes to mind very fast. This article is part of a series, start with Deepfake Fake Kiss Photo Of Georgia Mayor Outran The Law.

But a face bank has no opinion about guilt. It ranks lookalikes. A human has to do the rest: check the timeline, check the other evidence, and ask whether the person even could have been there.

The complaint alleges authorities knew AI facial recognition technology should not be used as the sole basis to extradite and charge a person, yet ignored those issues.

Lipps' lawsuit, as described by Fox News

The lawsuit also says the Fargo Police Department knew that failing to supervise how officers used the technology could lead to wrongful detention and prosecution, and did not supervise anyway. That is an allegation, not a ruling. But if it holds up, it is a policy problem, not a software glitch. A written policy (the rules officers must follow before acting on a match) is cheap. A missing one is expensive.

The Fix Already Exists: What Detroit Changed

Detroit's police settlement, covered by the American Bar Association, limits facial recognition to serious crimes like violent offenses or first-degree home invasions. Nobody had to invent anything new. The rules are out there. Fargo allegedly just didn't use them.

Is This a One-Off? Face Banks, Wrong Arrests, and the Rights of Ordinary People

It would be comforting to call Lipps' case a freak accident. The reporting says otherwise. Experts quoted in the coverage warn that police AI mistakes have already led to at least 16 wrongful arrests. A California man also sought $10 million after facial recognition falsely tied him to a Sunglass Hut robbery. He says he was sexually assaulted in jail. Two people, two lawsuits, the same dollar figure.

Then there's the bias problem. According to research discussed by the Bulletin of the Atomic Scientists, false positive rates (how often the system wrongly says two different people are the same) can be about two orders of magnitude higher for darker-skinned people. East African women produced roughly 100 times more false positives than white men. Your rights should not depend on whose face the software handles best.

And here's the twist that surprised me. Investigators are least likely to question a result when the image looks great. A crisp, clear photo feels trustworthy, so the red flags get waved off. A clean picture does not make a correct match. Previously in this series: Age Verification Supreme Court Eyes Liquor Shop Id Checks.

The Other Side: The Tool Isn't the Villain

Fair is fair. Policing Insight reports that top algorithms now show false positives around 0.3%, and that UK systems return the correct identity in 99% of cases. That's good. But a 0.3% error rate is still a real person once you search millions of faces. Better software lowers the odds. Only human checking catches the leftovers.

Anyone trying to educate consumers and celebrities about deepfakes and false matches would start in the same place: a lookalike is a lead, not a verdict. Lawyers dissecting the case at Law Commentary make the same point from the civil rights side.

What Can You Do If a Face Bank Match Points at Someone You Love?

If a computer match ever points at you or a family member, ask one question out loud: "What else, besides the match, says it's this person?" If the answer is "nothing yet," that is a lead being treated like a conclusion. Write down where you were, and gather proof like receipts, card charges, and location history. Lipps' own bank records did the heavy lifting. Do that early, not after six months.

If you've ever wondered whether a photo or profile is really the person it claims to be, that's the exact question comparison technology exists to answer. The responsible version produces a side-by-side comparison a human can inspect, with a clear note of what it can't tell you. A bare "match" with no explanation should make you nervous.

Here's what I can't stop thinking about. The person who finally cleared Angela Lipps wasn't an expert or an algorithm. It was a bank statement showing a woman buying ordinary things in Tennessee. That record existed the entire time. How many days of her life did it sit there, unread, while a computer's guess did the talking?

Face Bank: Frequently Asked Questions