On July 14, Tennessee grandmother Angela Lippswas arrested and spent more than five months in jail for a crime in North Dakota she did not commit. Her wrongful arrest stemmed from an AI facial recognition match, despite evidence showing she was not the person who committed the crime and had never even visited the state. Unfortunately, her story is not unique. Wrongful arrests tied to technology are happening across the country.
Robert Dillon, from Lee County, Florida, was arrested on a warrant out of Jacksonville—hundreds of miles away—accused of luring or enticing a child. Facial recognition technology had produced a 93% match to his photo. His image was even placed in a photo lineup for a witness two months later, who then identified him. But beyond the tech-generated match and the lineup, no evidence connected Dillon to the crime. His charges were eventually dropped and his record cleared, but not before his life was upended.
In another case in Orlando, Florida, a man was accused of fraud and theft tied to a hotel stay. The initial encounter was captured on body camera footage when the suspect was issued a trespass warning. Law enforcement believed the man gave a false identity, leading officers in another county to arrest Beau Burgess on those charges. Burgess repeatedly insisted he was innocent and provided a timecard proving he was at work—70 miles away—when the incident occurred.
When a local news outlet requested body camera footage and followed up with the Orlando Police Department, officials initially denied that facial recognition had been used. However, a later internal investigation revealed officers had used the FACES program, relying on a decades-old mugshot and placing Burgess in a photo lineup. A hotel employee then identified him. Ultimately, the state attorney dropped the case.
Orlando Police Department’s Policy on Facial Recognition states:
“Facial Recognition is an investigative tool and any law enforcement action taken based on a submission to FACESNXT or any other facial recognition system shall be based on the agency’s own identity determination and not solely the results of a facial recognition search. The result of a facial recognition search shall only be considered as an investigative lead and is not to be considered a positive identification of any subject or probable cause for arrest.”
During the 2026 Florida Legislative Session, lawmakers attempted to address these concerns. HB 875 / SB 1202 proposed:
“To include a suspect in a lineup, a law enforcement officer or agency must have an evidence-based reason to believe that such suspect committed the crime under investigation. If facial recognition technology is used to identify a suspect, a law enforcement officer or agency may not conduct a lineup unless there is a basis, independent of the use of facial recognition technology, to support a belief that the suspect committed the crime under investigation.”
Neither bill received a hearing.
AI has been immensely valuable in helping law enforcement prevent crime, protect communities, and reduce victimization. But to preserve both safety and civil liberties, guardrails are essential. Legislatures can provide clarity, accountability, and oversight by standardizing practices, requiring accurate reporting, maintaining inventories of technology in use, and ensuring proper auditing. These measures help lawmakers—and the public—understand how AI tools are being used in their communities while ensuring they are used responsibly and effectively.