A Tennessee grandmother spent over five months behind bars after a facial recognition tool used by North Dakota police mistakenly linked her to a multi-state bank fraud ring.
Angela Lipps, 50, who says she had never stepped foot in North Dakota, was released on Christmas Eve only after her defense attorney provided bank records proving she was more than 1,000 miles away when the crimes occurred.
The case has ignited a fierce debate over law enforcement’s growing reliance on artificial intelligence (AI) and “shortcuts” investigators take when high-tech tools provide a potential match.
The ordeal began on July 14, 2025, when Lipps was arrested in Tennessee on a fugitive warrant issued out of Fargo. Investigators in North Dakota had been searching for a suspect in a series of local bank thefts.
Lacking a lead, they turned to a “partner agency” in West Fargo that utilized Clearview AI, a controversial startup that identifies individuals by scraping billions of photos from social media and the open internet.
The AI flagged Lipps as a “potential suspect” based on similar facial features. While West Fargo police passed the lead to Fargo detectives, they noted they did not have enough evidence to file charges themselves. However, Fargo investigators moved forward with a warrant.
“It was the first time I had ever been on an airplane,” Lipps wrote in a verified GoFundMe post, describing her extradition. “I was terrified and exhausted and humiliated.”
Fargo Police Chief Dave Zibolski admitted to “a few errors” during a Tuesday press conference, specifically citing a lack of oversight regarding how neighboring departments use AI. He revealed that Fargo’s executive leadership was unaware West Fargo had even purchased the Clearview AI system.
“We would not have allowed that to be used, and it has since been prohibited,” Zibolski said, according to CNN. He explained that detectives “wrongly assumed” the AI report included sufficient corroborating evidence, such as surveillance photos, which were never actually cross-referenced with the state’s certified facial recognition center.
The case of Lipps has pulled back the curtain on a growing crisis in American policing: the “automation of probable cause.” Lipps, who had never set foot in North Dakota, was only exonerated after her attorney produced bank records proving she was 1,000 miles away during the fraud.
Lipps’ legal team has been scathing in their assessment of the investigation. In a statement to CNN, they argued that police ignored the most basic investigative steps.
“Officers knew that Angela was a Tennessee resident, and we have seen no investigation by officers to determine whether she traveled to or was in North Dakota… Instead, an officer used AI facial recognition as a shortcut.”
Despite the life-altering error, Chief Zibolski stopped short of a direct apology, noting the fraud case remains “open and active.” He blamed the lack of a “notification mechanism” between states for the months-long delay in Lipps’ extradition and interview.
Criminology experts warn that Lipps’ case is a symptom of a larger trend. Ian Adams, an assistant professor at the University of South Carolina, told CNN that agencies are adopting AI so rapidly they are forced to rely solely on “vendor promises” rather than proven efficacy.
“We get nightmare scenarios when we don’t have people doing what they’re supposed to do, with technology that they’re using inappropriately,” Adams said, noting that powerful algorithms can “lull” detectives into a dangerous sense of complacency.
Lipps, a mother of three and grandmother of five, is currently exploring civil rights claims. While she is back home in Tennessee, her attorneys say the “trauma and loss of liberty” cannot be easily repaired. As for North Dakota, Lipps was clear: “I’ll never go back.”
Facial recognition technology, often marketed as a high-precision forensic tool, is increasingly being used as an investigative shortcut. Experts point to a “Magic Box” fallacy, where detectives treat AI leads as absolute truth rather than preliminary suggestions. This reliance is compounded by documented algorithmic bias; NIST studies have repeatedly shown that facial recognition software is significantly less accurate when identifying women and people of color.
A chilling parallel is found in the 2023 case of Porcha Woodruff. The Detroit native was eight months pregnant when she was arrested at her home for robbery and carjacking — crimes she did not commit. Like Lipps, Woodruff was identified solely through a facial recognition match from an out-of-date photo. Despite the physical impossibility of a heavily pregnant woman committing the high-octane crimes described, the high-tech match outweighed common-sense evidence.
These incidents underscore a systemic failure in the investigative chain. While most departmental policies state that an AI match is not sufficient for an arrest warrant, the Lipps and Woodruff cases suggest that when the computer speaks, the human instinct for verification often falls silent.
