Six Months for a Crime She Didn't Commit: The AI Facial Recognition Error That Jailed Angela Lipps
A Tennessee grandmother was arrested at gunpoint, held without bail for nearly six months, and transported 1,200 miles to North Dakota — because an AI facial recognition system said she looked like a bank fraud suspect. She'd never been to North Dakota in her life. She lost her house, her car, and her dog. No one apologized.
What Happened
On July 14, 2025, a team of U.S. Marshals arrived at Angela Lipps' home in Elizabethton, Tennessee. She was 50 years old, a mother of three, a grandmother of five, and by her own account had spent virtually her entire life within the borders of north-central Tennessee and neighboring states. She had never been on an airplane.
The Marshals arrested her at gunpoint while she was babysitting four young children. She was booked into her county jail as a fugitive from justice — specifically, a fugitive from North Dakota, a state she says she had never visited.
The reason: months earlier, Fargo police investigators had been working a bank fraud case. A woman was seen on surveillance video using a fake U.S. Army military ID card to withdraw tens of thousands of dollars from banks in and around Fargo. Detectives ran the surveillance image through a facial recognition software system. The system returned a match: Angela Lipps.
According to court documents obtained by WDAY News through an open records request, a Fargo detective then compared Lipps' social media photos and Tennessee driver's license picture to the surveillance footage. He concluded in his charging document that Lipps "appeared to be the suspect based on facial features, body type and hairstyle and color." Four counts of unauthorized use of personal identifying information and four counts of theft were filed in Cass County, North Dakota.
At no point before the arrest did anyone from the Fargo Police Department call Angela Lipps to question her.
108 Days in a Tennessee Cell — Without Going to North Dakota
Because she was classified as a fugitive from another state, Lipps was held without bail. She sat in a Tennessee jail cell for 108 days — more than three and a half months — while North Dakota authorities made no effort to retrieve her.
North Dakota officers finally transported Lipps to Fargo on October 30. She appeared in a Cass County courtroom for the first time the following day, nearly four months after her arrest.
Her North Dakota attorney, Jay Greenwood, immediately requested her bank records. He obtained them. Investigators then met with Lipps and Greenwood at the Cass County jail on December 19. It was the first time Fargo police had interviewed her directly — more than five months after arresting her.
The bank records told a clear story. At the exact times police claimed Lipps was in Fargo committing fraud, her financial records showed she was in Tennessee — buying cigarettes at a gas station, ordering food through an Uber Eats account, depositing Social Security checks locally. She was 1,200 miles away.
"Around the same time she's depositing Social Security checks … she is buying cigarettes at a gas station, around the same time, she is buying a pizza, she is using a cash app to buy an Uber Eats," Greenwood told WDAY News.
On December 24, 2025 — Christmas Eve — the case was dismissed. Lipps was released from jail.
What She Lost
While Angela Lipps was in jail and unable to pay her bills, she lost her home. She lost her car. She lost her dog.
When she was finally released on Christmas Eve, Fargo police did not pay for her trip home. She was stranded. Local defense attorneys helped cover a hotel room and meals on Christmas Eve and Christmas Day. A local nonprofit, the F5 Project, helped fund her return to Tennessee.
No one from the Fargo Police Department has apologized, according to Lipps. The department acknowledged "a few errors" in the case and pledged changes to its operations, but stopped short of a direct apology, CNN reported.
Lipps' attorney Greenwood told local outlet Valley News Live that "the cloud of suspicion has not been fully lifted" from his client, even after the charges were dismissed.
A lawsuit is being planned, according to FindLaw, which reviewed court documents in the case.
How the Technology Failed
Facial recognition systems work by converting a face — from a photo, surveillance frame, or ID — into a mathematical representation, then comparing it against a database to find the closest match. The software does not determine innocence or guilt. It identifies candidates. Human investigators are then supposed to confirm or rule out a match through independent investigation.
In the Lipps case, investigators used the facial recognition output as the foundation for an arrest warrant. The detective's own charging document acknowledged that the identification relied on the AI result plus a visual comparison of social media photos and a driver's license — not bank records, alibi investigation, or any attempt to contact Lipps before filing charges.
"If the only thing you have is facial recognition, I might want to dig a little deeper," her attorney Greenwood said.
The Fargo case is not an isolated failure. The technology has a documented bias problem that has produced a pattern of wrongful arrests. According to the Innocence Project, there are at least seven confirmed cases in the United States of individuals wrongfully arrested based on facial recognition misidentification. Six of those seven cases involve Black individuals.
A 2019 federal study by the National Institute of Standards and Technology found that Asian and Black people were between 10 and 100 times more likely to be misidentified by facial recognition systems than white people. Older systems tested even worse. The accuracy gap is most severe with lower-quality images — such as surveillance camera footage, the exact type of image used in the Lipps case.
A 2025 Washington Post investigation found that of the wrongful arrests linked to facial recognition across the U.S., all seven additional wrongful arrestees identified by the Post were Black.
A Pattern of Documented Failures
The Lipps case arrives amid a string of high-profile facial recognition failures in the United States and United Kingdom.
In October 2025, an AI system at a Baltimore high school apparently identified a bag of Doritos as a firearm, triggering a police response. Officers arrived with guns drawn, forced a student named Taki Allen to kneel, and handcuffed him. Nothing was found.
In the United Kingdom, police arrested a man for a burglary in a city he had never visited after facial recognition confused him with another person of South Asian heritage. The suspect had been in a city more than 100 miles from the crime scene.
Robert Williams, a Black Detroit man, was arrested in front of his family in 2020 after facial recognition incorrectly identified him as a shoplifting suspect. He spent 30 hours in custody before the case collapsed. The city of Detroit later agreed to a landmark settlement — including an audit of all cases since 2017 in which facial recognition was used to obtain an arrest warrant — following litigation covered extensively by University of Michigan Law School's Law Quadrangle publication.
Porcha Woodruff, also of Detroit and eight months pregnant, was arrested in 2023 on a carjacking and robbery charge traced to a facial recognition match. The charges were dropped. She filed a lawsuit.
Nijeer Parks of New Jersey was jailed for ten days in 2019 after facial recognition linked him to a crime he did not commit. He spent nearly $5,000 in legal fees defending himself. His lawsuit against the police department was settled.
Regulation and Its Absence
There is no federal law in the United States governing how police may use facial recognition technology, the standards required before it can form the basis of an arrest warrant, or the procedural safeguards required afterward.
A small number of jurisdictions have enacted local restrictions. San Francisco, Boston, and Portland, Oregon have banned government use of facial recognition. Several other cities have imposed limits. But most police departments in the United States operate under no binding rules — either on the technical standards the systems must meet or on the investigative steps required after a facial recognition match is returned.
Illinois' Biometric Information Privacy Act, one of the strongest biometric data laws in the country, regulates private companies' collection and use of facial data — but does not govern law enforcement use.
Multiple federal bills addressing law enforcement use of facial recognition have been introduced in Congress in recent years. As of March 2026, none have been enacted.
The Department of Justice has voluntary guidelines recommending that facial recognition alone should not form the sole basis for an arrest warrant. The Fargo Police Department acknowledged in its statement to CNN that investigators had taken "additional investigative steps independent of AI to assist in identification" — but those steps did not include contacting Lipps, checking her bank records, or confirming she had ever been to North Dakota before a warrant was issued and federal marshals deployed.
What the Numbers Show
According to a 2025 survey of law enforcement agencies by the Government Accountability Office, more than half of federal law enforcement agencies use facial recognition technology — including Customs and Border Protection, the FBI, and Immigration and Customs Enforcement. State and local police use of the technology is widespread but largely untracked at the federal level.
The NIST 2019 testing — the most comprehensive government evaluation of facial recognition accuracy — tested 189 algorithms from 99 developers. The findings on racial and demographic disparities were unambiguous: the highest false positive rates were consistently found among Black women, East Asian faces, and images of older individuals. The exact demographic profile most likely to be misidentified — a middle-aged woman photographed from a surveillance camera at a distance — is the profile of Angela Lipps.
NIST has continued updating its testing. Accuracy across all systems has improved. The disparity between demographic groups has not been eliminated.
Where the Case Stands
Angela Lipps is back in Tennessee. The charges against her have been dismissed. She is working to rebuild what she lost — the house, the vehicle, the normalcy of daily life.
The woman who actually committed the bank fraud in Fargo — the person seen on the surveillance video using a fake military ID — has not been publicly identified or charged, according to available reporting through the date of this article.
The Fargo Police Department has said it is reviewing its facial recognition protocols. It has not said what specific changes will be made, on what timeline, or whether any review of past cases is planned.
No one has apologized to Angela Lipps.