Wrongful Arrest of Randal Reid Based on Facial Recognition Error in Louisiana

In August 2022, Randal Reid, a 28-year-old Black man from Georgia, was arrested in DeKalb County, Georgia, on an outstanding warrant from Jefferson Parish, Louisiana. The warrant was issued based on a facial recognition match that allegedly connected Reid to a theft at a Sunglass Hut store in Metairie, Louisiana, a city Reid had never visited. The theft occurred on May 25, 2022, but Reid was not arrested until late August 2022. The Jefferson Parish Sheriff's Office used facial recognition technology to analyze surveillance footage from the theft and generated a match to Reid's driver's license photo. Acting solely on this algorithmic match, investigators obtained an arrest warrant without conducting additional verification or investigating Reid's whereabouts during the alleged crime. Reid was subsequently arrested at his home in Georgia and extradited to Louisiana, where he spent six days in Jefferson Parish Correctional Center. During his incarceration, Reid's attorney and family worked to establish his innocence by providing evidence that he was in Georgia at the time of the Louisiana theft. This included employment records, cell phone location data, and witness testimony that contradicted the possibility of his presence in Louisiana during the crime. The charges against Reid were eventually dropped when prosecutors acknowledged they could not prove he was in Louisiana at the time of the theft. This case represents a growing pattern of wrongful arrests based on facial recognition technology, which has documented higher error rates for Black individuals due to algorithmic bias. The American Civil Liberties Union has represented multiple similar cases, including those of Robert Julian-Borchak Williams and Nijeer Parks, highlighting systemic issues with law enforcement's reliance on facial recognition without adequate safeguards. Reid's case prompted renewed calls for restrictions on facial recognition use in policing and mandatory human verification protocols before arrests can be made based solely on algorithmic matches.