Facial Recognition Named the Wrong Person, and Nobody Checked for Six Months

What happened

On July 14, 2025, Angela Lipps was at her home in Tennessee babysitting her grandchildren when police arrived and arrested her at gunpoint. She was wanted in North Dakota on bank fraud charges she had nothing to do with. A facial recognition system operated by the West Fargo Police Department had matched her face to a suspect in a series of crimes that took place in Fargo and West Fargo between April and May 2025. A North Dakota judge signed her arrest warrant on July 1 based on that match and a cursory comparison of social media photos.

The fraud had been real. Someone had been stealing from bank customers in North Dakota. But the investigation that followed the AI match appeared to stop there. Detectives did not verify whether Lipps could have been in North Dakota during the crimes before seeking a warrant. She was arrested, extradited to North Dakota on October 30, and held in the Cass County jail for nearly six months. During that time she lost her home, her car, and her dog, and suffered what the record describes as serious trauma and reputational damage.

The exoneration came on December 19, 2025, when Lipps' attorney presented bank records at the jail. The records showed she had been in Tennessee throughout the entire period when the fraud was committed, more than 1,200 miles from North Dakota, buying cigarettes at a gas station, depositing Social Security checks, ordering pizza, and using a cash app for food deliveries. The charges were dismissed on December 24 and she was released. The alibi was not constructed after the fact. It had been sitting in ordinary financial records the whole time.

The cause was a false positive from the facial recognition tool combined with what the record describes as a failure of investigative verification. Automation bias, the tendency to treat an algorithmic output as a conclusion rather than a lead, meant the match was treated as sufficient grounds for an arrest warrant without the geographic and financial checks that would have cleared Lipps quickly. The record notes that without binding national standards on how facial recognition evidence must be corroborated before it reaches a judge, this pattern will repeat.

What the case leaves visible is the absence of any required step between an AI match and a court-endorsed arrest. Nobody had to document what the system produced, what confidence score accompanied the result, or what human review took place before the warrant was signed. A provable record of what a system did, when a reviewer examined it, and what verification followed would not have prevented the false positive, but it would have made the gap between the match and any corroborating evidence impossible to ignore before the warrant was signed, not six months after.

Reported impact

Affected parties

Not publicly disclosed

Harm type

Not publicly disclosed

Scale

Not publicly disclosed

Financial impact

Not publicly disclosed

Regulatory action

Not publicly disclosed

Classification

Organization

Not publicly disclosed

AI system

Not publicly disclosed

Industry

Not publicly disclosed

Country

Not publicly disclosed

Provider

Not publicly disclosed

Incident type

Not publicly disclosed

Relevant governance controls

Governance control mapping is not available for this record.

• No controls mappedNot publicly disclosed

Control mapping is analytical. It does not state that any control would have prevented the incident.

Sources and evidence

This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.