TL;DR: A Washington Post investigation found at least 8 Americans wrongfully arrested after facial recognition AI identified them as suspects, with no other evidence. Of the 8 known cases, 7 are Black. Police in 15 departments across 12 states arrested people based solely on AI matches, violating their own policies. One man spent over a year in jail. States are now passing laws requiring corroborating evidence, but most departments still have no guardrails.
What changed (September 16, 2026): The ACLU's April 14, 2026 update identified "more than a dozen" documented wrongful facial recognition arrests, adding six cases to the original eight (Michael Oliver in Detroit, Alonzo Sawyer in Maryland, Robert Dillion in Jacksonville Beach, Javier Lorenzano-Nunez in Phoenix, Beau Burgess in Orlando, and Kimberlee Williams in Oklahoma). The featured new case, Kimberlee Williams, was held for six months after Maryland police used facial recognition to identify her in a case where she had never been to Maryland.
The Algorithm Said He Did It
Christopher Gatlin spent over a year in jail awaiting trial. The charge: a crime he didn't commit. The evidence: a facial recognition match.
That's it. A computer said his face looked like the suspect's face. Police arrested him. Prosecutors charged him. The justice system held him for more than 365 days before the case was dropped.
Gatlin isn't alone. The Washington Post reviewed documents from 23 police departments where detailed records about facial recognition use are available. The findings: 15 departments in 12 states arrested suspects identified through AI matches with zero independent evidence connecting them to the crime.
No fingerprints. No eyewitnesses. No DNA. Just a software suggestion treated as gospel.
The Cases Keep Piling Up
Jason Vernau spent three days in a Miami jail after police accused him of cashing a $36,000 fraudulent check. The facial recognition system matched him to bank surveillance footage. Police never checked his bank accounts. If they had, they'd have seen he had no connection to the fraud. Wrong guy.
Porcha Woodruff was seven months pregnant when Detroit police arrested her for a carjacking. The surveillance footage showed no pregnant woman. Eyewitness statements mentioned no pregnant woman. The facial recognition system matched her face, and that was enough.
Trevis Williams got jailed for two days by the NYPD on a false match. He was 8 inches shorter and 70 pounds lighter than the actual suspect. The only similarity: both were Black men with locks. Civil rights groups are now demanding an investigation into the NYPD's use of the technology.
Jason Killinger was arrested in September 2023 at the Peppermill casino in Reno, Nevada. The casino's facial recognition flagged him as a previous trespasser. The two men have different eye colors: Killinger's are blue, the other man's hazel. The arresting officer claimed these are "by their very nature similar eye colors and are dependent upon lighting." Killinger is now suing for Fourteenth Amendment due process violations.
Of the eight confirmed wrongful arrests documented in the Post investigation, seven of the victims are Black.
Departments Violate Their Own Rules
Most police departments have written policies requiring officers to corroborate facial recognition matches before making arrests. The Post found officers routinely ignore them.
One police report referred to an uncorroborated AI result as a "100% match." Another said police used the software to "immediately and unquestionably" identify a suspected thief. No such certainty exists in facial recognition technology. These systems generate probability scores, not fingerprints.
Studies consistently show facial recognition misidentifies Black people and other people of color at higher rates than white people. A UK Home Office study found that Black and Asian people are more likely to be incorrectly matched than white people. Yet police treat every match as equally reliable.
The EFF put it bluntly: "Police say a simple warning will prevent face recognition wrongful arrests. That's just not true."
States Start Adding Guardrails
Some states are finally acting. New laws prohibit police from going straight from facial recognition results to witness identification procedures. Officers cannot apply for arrest warrants based solely on an AI match.
But these laws are exceptions, not the rule. Most of the 18,000+ law enforcement agencies in America have no facial recognition policies at all. No oversight. No audits. No accountability when the technology fails.
Meanwhile, departments are expanding use. Dallas police used Clearview AI 156 times since adoption, contributing to 25 arrests. Now they're considering expanding the technology to lesser crimes like trespassing and package theft. The inevitable result: more matches, more arrests, more mistakes.
What You Can Do
If You're Arrested on a Facial Recognition Match
Demand to know if facial recognition was used in your case. Request all documents related to the match, including the probability score. Contact the ACLU or a civil rights attorney immediately. Document your alibi: if you can prove you were elsewhere, the case collapses.
Protect Yourself Preemptively
Limit photos on social media. Clearview AI scrapes public photos. Support local bans on police facial recognition. Learn techniques to defeat facial recognition. Keep timestamped records of your whereabouts (receipts, photos with metadata, location history).
Push for Policy Changes
Demand your city council ban or regulate police facial recognition. Support state legislation requiring corroborating evidence for arrests. Push for transparency. Departments should publish how often they use this technology and what happens after matches. File public records requests.
When Algorithms Replace Investigation
Facial recognition was sold to police as a tool to help investigations. Instead, it's replacing them. Why interview witnesses when the computer already picked someone? Why check alibis when the AI is confident?
Every wrongful arrest represents a double failure: an innocent person jailed, and the actual perpetrator walking free. When Porcha Woodruff sat in a Detroit cell, someone who actually committed a carjacking went home.
The technology isn't going away. Dallas is expanding. New Orleans has live facial recognition networks. The NYPD uses it routinely. The question is whether we'll require the same evidentiary standards we've always required, or let an algorithm convict.
Christopher Gatlin lost a year of his life because a computer said he looked guilty. How many more will pay that price before we demand better?
Update (September 16, 2026)
The "at least 8" framing this article opened with is now conservative. On April 14, 2026, the ACLU's Speech, Privacy, and Technology Project published an update titled "More than a Dozen Wrongful Arrests Due to Police Reliance on Facial Recognition Technology," which added six documented cases to the original Washington Post eight.[7] The new additions are:
- Michael Oliver (July 2019, Detroit, MI): Arrested after a facial recognition match; charges later dropped. A separate but adjacent Robert Williams case in Detroit in January 2020 became the first publicly litigated wrongful facial recognition arrest.
- Alonzo Sawyer (March 2022, Maryland transit police): Identified through facial recognition in a case where the underlying investigative steps were not independently corroborated.
- Robert Dillion (August 2024, Jacksonville Beach, FL): Wrongful arrest on a facial recognition match, one of two Florida cases added in the past year.
- Javier Lorenzano-Nunez (October 2024, Phoenix, AZ): Arrested after a Phoenix Police Department facial recognition query produced a candidate match without other evidence.
- Beau Burgess (August 2025, Orlando, FL): Florida's second recent addition.
- Kimberlee Williams (featured; arrested June 23, 2021, on Maryland warrants while living in Oklahoma): Montgomery County, Maryland police used facial recognition to identify her in an identity-theft case. Williams had never been to Maryland. She spent roughly six months in jail before charges were dropped in December 2021. The Washington Post profiled her case in early 2026 and the ACLU's Maryland affiliate has been pushing the department for an apology.
Two further developments in the same window are worth flagging for the record. First, the federal lawsuit brought by Porcha Woodruff (named in this article) against the City of Detroit has been resolved in 2025; the case was the most prominent test of municipal liability for a facial recognition wrongful arrest, and its disposition is the reason civil rights attorneys are now pushing state-level corroboration statutes rather than relying on §1983 actions alone. Second, the August 2024 ACLU lawsuit against the Clearwater Police Department and the Pinellas County Sheriff's Office, filed on behalf of a Florida man arrested on an 85%-accuracy facial recognition match, remains the most prominent case in which the algorithm's reported confidence score is itself part of the pleading.[8]
Two patterns this article identified in December 2025 have hardened since. Every new case in the ACLU update involves a Black defendant or a defendant who would not have been identified at all without the AI match, and in every new case the arresting department skipped the corroboration step its own written policy required. The 18,000+ law enforcement agencies without facial recognition policies remain the larger structural exposure; the named cases are the visible surface of an invisible pipeline.
Sources for this update: ACLU. "More than a Dozen Wrongful Arrests Due to Police Reliance on Facial Recognition Technology." April 14, 2026; The Daily Record. "Facial recognition error led to woman jailed six months in MD." April 16, 2026; The Guardian. "ACLU sues on behalf of Florida man wrongfully arrested after facial recognition error." August 13, 2024.
References
- Washington Post - Arrested by AI: Police ignore standards after facial recognition matches (2025)
- ABC7 New York - Man's wrongful arrest puts NYPD's use of facial recognition under scrutiny (December 2025)
- Casino Beats - Reno Police Report Shows How Casino's Facial Recognition Led to Wrongful Arrest (December 9, 2025)
- Electronic Frontier Foundation - Police Use of Face Recognition Continues to Wrack Up Real-World Harms (January 2025)
- ACLU - Police Say a Simple Warning Will Prevent Face Recognition Wrongful Arrests. That's Just Not True
- Stateline - Facial recognition in policing is getting state-by-state guardrails (February 2025)
- Dallas Morning News - Dallas police use AI face recognition tool in serious felonies. They may add lesser crimes (December 8, 2025)
- ACLU. "More than a Dozen Wrongful Arrests Due to Police Reliance on Facial Recognition Technology." April 14, 2026. By Lauren Yu and Nathan Freed Wessler (ACLU Speech, Privacy, and Technology Project). Adds Michael Oliver (Detroit, July 2019), Alonzo Sawyer (Maryland transit police, March 2022), Robert Dillion (Jacksonville Beach, August 2024), Javier Lorenzano-Nunez (Phoenix, October 2024), Beau Burgess (Orlando, August 2025), and Kimberlee Williams (Oklahoma/Maryland, June 2021) to the original eight documented cases.
- The Guardian. "ACLU sues on behalf of Florida man wrongfully arrested after facial recognition error." August 13, 2024. Documents the 85% confidence score in the lawsuit pleading, the most prominent case in which the algorithm's reported confidence is itself part of the case.