She Delivered DoorDash. Maryland Jailed Her for Six Months.

Kimberlee Williams is the 14th person wrongfully arrested by facial recognition. Police hid the technology from the court. The ACLU is demanding answers.

State of Surveillance Editorial Team

Wooden judge gavel resting on a dark surface in a courtroom
Photo via Unsplash

TL;DR: Kimberlee Williams, a mother of seven working at a methadone clinic in Lawton, Oklahoma, was arrested on June 23, 2021 while on a DoorDash delivery with her daughter. The charge: bank fraud in Maryland, a state she'd never set foot in. An anonymous person on a law enforcement listserv called CrimeDex ran a bank surveillance photo through facial recognition software and flagged Williams as the suspect. Montgomery County Detective Michael Adami obtained a warrant without telling the court that facial recognition was involved. Williams spent 23 days in an Oklahoma jail, over three months in Montgomery County, and two more months in Prince George's County: six months total. She faced 16 charges across three counties, including 12 felonies. Every charge was eventually dismissed. The ACLU sent letters to all three police departments on April 14, 2026 demanding apologies and policy changes. Williams is the 14th known wrongful facial recognition arrest in the U.S. Nearly every victim is Black.

Arrested at a Military Gate

On June 23, 2021, Kimberlee Williams was riding with her 23-year-old daughter Jordyn Wasko on a DoorDash delivery to Fort Sill in Lawton, Oklahoma. Routine. Base security checked her ID. That's when they found the Maryland warrants [1].

Williams had never been to Maryland.

"I thought they were joking," Williams told the Washington Post. "I wasn't even scared, because I was like, 'I've never been to Maryland'" [1].

The warrants came from three different Maryland counties: Montgomery, Prince George's, and Anne Arundel. Someone had walked into SunTrust and Truist bank branches in December 2019, impersonated account holders, and withdrawn roughly $17,000 using forged checks. The suspect wore a red jacket on the first visit, changed her hairstyle and outfit for the second [1].

Williams was in Oklahoma for all of it. She was home with her family celebrating Christmas and her daughter's birthday while someone she'd never met was cashing fake checks 1,400 miles away [2].

The CrimeDex Pipeline

Here's how an Oklahoma grandmother ended up in Maryland handcuffs for crimes she didn't commit.

A SunTrust bank investigator grabbed the surveillance footage of the unknown suspect and posted it to CrimeDex, an online community where law enforcement officers and corporate investigators share bulletins and solicit tips. Think of it as a police Facebook group for identifying suspects [2].

Someone on CrimeDex responded. According to the bank investigator, a respondent "suggested, using facial recognition software" that Williams was the suspect [1]. Nobody knows who ran the search. Nobody knows what software they used. Nobody knows what database they searched against.

CrimeDex founder James Hudson acknowledged that "another CrimeDex user could have performed a facial recognition search from the bank investigator's bulletin and shared the results" [1]. That's it. An anonymous person ran an anonymous algorithm against an unknown database and produced a name. That name became a warrant. That warrant became six months in jail.

Detective Adami's Warrant

Montgomery County Detective Michael Adami took the CrimeDex tip and applied for an arrest warrant. He didn't mention facial recognition. Not once [2].

Instead, Adami wrote that Williams had been "identified" as the suspect. He claimed he'd confirmed the identification by visually comparing a surveillance photo to an older photo of Williams [2]. According to Iowa State psychology professor Gary Wells, who reviewed the photos for the Washington Post, the comparison was "egregious," noting that "these are very clearly different people" [1].

The bank investigator's emails to Prince George's and Anne Arundel County police also failed to disclose that facial recognition was used to identify Williams [2].

The ACLU's complaint is specific: Adami "concealed that the entire basis of Ms. Williams' purported identification was a lead from an unknown and unreliable face recognition technology search by an unknown entity" [2]. He hid the technology from the court. He hid it from the defense. And three counties ran with it.

Six Months, Three Jails

Williams sat in an Oklahoma jail for 23 days waiting for Maryland officers to pick her up. They transferred her to Montgomery County, where she spent over three months fighting to prove she couldn't have committed crimes in a state she'd never visited. Then Prince George's County took her for two more months [1].

During a Montgomery County interrogation in July 2021, Williams tried to explain. "I'm not trying to waste your time. I'm telling you..." An officer cut her off: "You're telling me, but you're not telling me the truth" [1].

Six months. Sixteen charges across three counties. Twelve felonies.

By December 2021, all charges were dismissed. The evidence against her was a facial recognition match from someone nobody can identify, confirmed by a detective's eyeball comparison that a psychology professor called obviously wrong [1].

What She Lost

Before the arrest, Williams had rebuilt her life. She'd had prior convictions for writing bad checks: misdemeanors in 2001 and 2004, a felony in 2010 that put her in prison for about two years. But by 2021, she was working as an administrator at a methadone clinic in Lawton. Raising her kids. Making DoorDash deliveries on the side [1].

"We were doing so much better," her daughter Jordyn said [1].

Williams lost her job. She lost months with her seven children. "She lost a part of herself," Jordyn told the Post [1].

As of 2025, Williams was living in a care facility in Hermitage, Pennsylvania, recovering from Guillain-Barré syndrome [1]. Her records have been expunged. No police department has apologized.

"I lost six months of my life when Maryland police wrongfully imprisoned me halfway across the country from my children, my home, and my job, all because they relied on an incorrect result from faulty technology," Williams said [2].

The 14th Name on a Growing List

Williams is the 14th person publicly known to have been wrongfully arrested by U.S. police because of facial recognition. The ACLU published the updated list on April 14, 2026 [3]:

  1. Nijeer Parks: Woodbridge, New Jersey, 2019
  2. Michael Oliver: Detroit, 2019
  3. Robert Williams: Detroit, 2020
  4. Christopher Gatlin: St. Louis, 2021
  5. Alonzo Sawyer: Maryland, 2022
  6. Randal Quran Reid: Jefferson Parish, Louisiana, 2022
  7. Porcha Woodruff: Detroit, 2023
  8. Jason Killinger: Reno, Nevada, 2023
  9. Robert Dillon: Jacksonville Beach, Florida, 2024
  10. Javier Lorenzano-Nunez: Phoenix, 2024
  11. Trevis Williams: New York City, 2025
  12. Angela Lipps: Fargo, North Dakota, 2025
  13. Beau Burgess: Orlando, 2025
  14. Kimberlee Williams: Maryland, 2021 (disclosed 2026)

Nearly every victim is Black. Williams is Black. The pattern holds because the technology fails on darker skin tones at rates 10 to 100 times higher than on white faces, according to NIST testing [4].

Maryland Changed Its Law. Too Late for Williams.

In 2024, Maryland passed legislation prohibiting police from using facial recognition as the sole basis for an arrest [1]. It was the kind of reform that might have prevented Williams' case, if it had existed three years earlier.

But the ACLU argues the law doesn't go far enough. Their demands to the three police departments include [2]:

  • A public apology to Williams
  • A ban on police relying on facial recognition searches conducted by outside entities (like anonymous CrimeDex users)
  • A prohibition on arrests based solely on a facial recognition match followed by a human visual comparison
  • Mandatory disclosure of facial recognition use in all warrant applications and court proceedings

Montgomery County, Prince George's County, and Anne Arundel County have not responded to the ACLU's letters [2].

The Same Playbook, 14 Times

Williams' case follows the exact playbook we've documented across all 14 wrongful arrests:

Step 1: An algorithm produces a match. The algorithm's accuracy, the database it searched, and sometimes even the identity of who ran it are unknown.

Step 2: A detective "confirms" the match by looking at two photos and deciding they look similar enough. Psychology researchers call this confirmation bias [1].

Step 3: Nobody checks whether the suspect was actually in the area. Nobody pulls location data. Nobody requests alibi verification.

Step 4: The warrant application hides the role of facial recognition. In Williams' case, it was called an "identification." In Randal Quran Reid's case, it was attributed to a "credible source" [5]. The defendant and the judge never learn a machine was involved.

Step 5: An innocent person goes to jail.

Fourteen times. Same steps. Same outcome. And those are only the cases where someone eventually figured out what happened.

What You Can Do

Demand Disclosure Laws

Only a few states require police to tell courts when facial recognition was used. Contact your state legislators and push for mandatory disclosure in all warrant applications. Maryland's 2024 law is a start, but Williams' case shows it came too late and doesn't cover outside searches.

Know That Warrants Can Hide It

If you're ever charged with a crime you didn't commit, ask your attorney to specifically request discovery on whether facial recognition, CrimeDex, or any automated identification tool was used. Police regularly disguise it behind phrases like "identified" or "credible source."

Support the ACLU's Push

The ACLU is demanding that police departments ban reliance on facial recognition searches conducted by outside entities and prohibit arrests based solely on algorithmic matches. Follow the case and support these reforms.

Limit Your Facial Data

CrimeDex users and police can run your photos through facial recognition without your knowledge. Minimize public-facing photos on social media. Opt out of data broker sites. The less your face is in searchable databases, the smaller the chance of a false match.

References

  1. Washington Post: 'That wasn't me': How facial recognition led to a woman being jailed for 6 months (April 14, 2026)
  2. ACLU: Woman Wrongly Jailed for Months Based on Faulty Facial Recognition Technology Demands Apology from Maryland Police Departments (April 14, 2026)
  3. ACLU: More than a Dozen Wrongful Arrests Due to Police Reliance on Facial Recognition Technology (April 2026)
  4. NIST: Face Recognition Vendor Test (FRVT)
  5. Seattle Times: Police relied on hidden technology and put the wrong person in jail
  6. ACLU: Kimberlee Williams Wrongful Arrest Case Page

Published: April 15, 2026