Facial Recognition Fail: Detroit Forced To Pay $300K To Man AI Falsely Recognised As Thief

The City of Detroit has agreed to pay a man $300,000 in a settlement after the city police’s facial recognition software wrongly identified him and accused him of shoplifting. The city will also revise its policies on how law enforcement uses facial recognition technology to solve crimes.

The lawsuit settlement involving citizen Robert Williams came from a 2018 incident where his driver’s license photo was mistakenly identified as a match to a suspect in security footage from a Shinola watch store.

“We are extremely excited that going forward there will be more safeguards on the use of this technology with our hope being to live in a better world because of it,” Williams told reporters, “even though what we would like for them to do is not use it at all.”

Accuracy And Bias In Facial Recognition

The American Civil Liberties Union (ACLU) and the Civil Rights Litigation Initiative (CRLI) at the University of Michigan Law School jointly announced the settlement agreement on Friday. They maintain that facial recognition technology is unreliable and biased against people of colour, particularly since Williams, the impacted individual, is Black.

Further highlighting the impact of his wrongful arrest, the ACLU pointed out that Williams was arrested outside his home, in front of his two young daughters and wife and in plain view of his neighbours, and subjected to thirty hours of detention in an overcrowded, dirty cell.

The ACLU reported that the Detroit police will be restricted from making arrests solely on facial recognition matches. Additionally, they cannot conduct arrests based on photo lineups derived from facial recognition searches.

“They can get a facial recognition lead, and then they can go out and do old-fashioned police work and see if there’s actually any reason to believe that the person who was identified … might have committed a crime,” said Phil Mayor, an ACLU attorney.

Detroit police haven’t commented on the settlement yet. Last August, during the lawsuit, Chief James White implemented new policies regarding the technology. This followed a separate incident where a woman, eight months pregnant, was mistakenly charged with carjacking.

According to Chief White’s previous statement, for police to utilise facial recognition, they must have additional evidence, independent of the technology, that suggests the suspect has the “means, ability, and opportunity” to commit the crime.

The Impact Of Misidentification

The settlement with Williams stipulates that Detroit police will conduct a retrospective review of cases between 2017 and 2023 where facial recognition was employed. If the review reveals an arrest made solely on facial recognition without additional evidence, the prosecutor will be notified.

Even though technology offers significant advantages in criminal investigations, its limitations can’t be ignored. The case of Robert Williams, who was wrongfully arrested due to facial recognition errors, exemplifies this perfectly. His situation isn’t an isolated incident.

In May, a 19-year-old woman, similar to the case of Williams, was misidentified as a shoplifter by Facewatch, a major biometric security company operating in the UK. Last month, OpenAI’s AI-powered chatbot, ChatGPT, falsely accused Jonathan Turley, a law professor at George Washington University, of sexually harassing a student.

Incidents like these raise concerns about the accuracy of facial recognition technology. Further development and regulations are needed to ensure this technology is used fairly and effectively.