This post is also available in:
עברית (Hebrew)
A police department in Detroit announced it will no longer be using facial recognition technology as the basis for arrest after wrongfully jailing the wrong person for 30 hours.
Facial recognition systems use biometrics to identify a person’s features from a video or photo, and then run the result through a database to verify the person’s identity. While this kind of system is widely used in many public areas, there have been several instances of wrongful identification of uninvolved individuals as suspects, leading to unjust arrests.
Back in 2020, Robert Williams spent 30 hours in jail after facial recognition technology had wrongfully declared him a shoplifter based on security camera footage. This reportedly ‘grainy’ footage was then run on the police system and compared to driver’s license photos and mugshots of over 243 individuals.
Williams’ old driver’s license photo was ninth on the list and was flagged by the operator as the best match, sent to the police department, and presented amongst five other individuals in a photo lineup. The store’s security contractor was then asked to confirm the suspect who was ‘closest’ to the accused they saw, after which a warrant was issued for Williams’ arrest, who spent the night in jail and had to hire a lawyer. After being released and proving he was at his workplace at the time of the crime, Williams sued Detroit in 2021 with the aim of ending the use of the technology that led to his wrongful arrest.
According to Interesting Engineering, last week the Detroit police department reached a settlement with Williams in which they provided monetary compensation and promised to not use facial recognition technology as the sole basis for making arrests. They further committed to not conduct “lineups” without reliable evidence linking a suspect to the crime or based solely on facial recognition results.
