In February 2023, Porcha Woodruff—a 32-year-old Black woman who was 8 months pregnant—was arrested at her Detroit home for robbery and carjacking. The evidence? A faulty facial recognition match.
The Arrest
Six police officers arrived before 8 a.m.
Her two children (ages 12 and 6) watched
She was detained for 11 hours
She experienced contractions in custody
Required hospitalization after release
The Problem: The surveillance footage showed no indication the carjacker was pregnant. The algorithm simply got it wrong.
Watch this investigation into how facial recognition bias leads to wrongful arrests:
The Pattern
Woodruff was the first woman wrongfully arrested by facial recognition. She was not the first person.
Known Wrongful Arrests (2020-2024):
Robert Williams (Detroit, 2020) - Watch theft, detained 30 hours
Michael Oliver (Detroit, 2019) - Also wrongfully arrested
At least 7 documented cases nationwide (ACLU, April 2024)
Nearly all victims were Black — in nearly every known case, the person wrongfully arrested was Black (ACLU, April 2024)
The Racial Bias
Facial recognition systems misidentify Black people at significantly higher rates than white people. Studies show:
5-10x higher error rates for Black people in NIST government testing (WIRED, July 2019)
Female faces misidentified more than male
Algorithm training data is predominantly white faces
The Broken Process
Police departments routinely:
Ignore warnings that FRT results aren't "positive identification"
Move directly from algorithm match to arrest warrant
Fail to conduct independent corroborating investigation
Omit FRT use from warrant applications
The Detroit Settlement (2026)
After Robert Williams' lawsuit, Detroit implemented the nation's strictest FRT policies:
Cannot go directly from FRT match to witness identification
Cannot obtain arrest warrants based solely on FRT results
Must conduct independent corroborating investigation
Making Change
Support organizations like ACLU and EFF
Advocate for biometric privacy legislation
Know your rights if approached by police
Document any use of facial recognition in your encounters
The Question: The ACLU's assessment is blunt: police departments frequently use face recognition to identify unknown suspects by comparing images to large photo databases, and it often fails to generate a correct match. In nearly every known case, the person wrongfully arrested was Black.