Detroit Police AI Software Policies Pivot Amid Wrongful 2020 Arrest Of Black Man

Robert Williams spent 30 hours in jail for a crime he did not commit.

As part of a federal settlement, the Detroit Police Department has overhauled its facial recognition software policies following the wrongful arrest of Robert Williams in 2020.

The pivot comes after Williams, a Black man, was mistakenly identified as a theft suspect due to flawed technology implementation. According to the Detroit Free Press, the agreement, filed in U.S. District Court, includes a $300,000 compensation for Williams, approved by the Detroit City Council in May.

BLACK ENTERPRISE previously covered the 2018 incident, which resulted in Williams being arrested in 2020 in front of his family for a Shinola store robbery he didn’t commit. A subpar image from security footage led a detective to misidentify Williams based on a false match with the man’s expired driver’s license photo, resulting in his 30-hour detention. Following the dismissal of his case without prejudice, Williams took legal action against the Detroit Police Department, citing the distressing ordeal inflicted upon him and his loved ones.

During a press conference on June 28, the American Civil Liberties Union (ACLU) of Michigan, representing Williams, unveiled sweeping changes to Detroit police software policies. These new guidelines prohibit arrests based solely on facial recognition results or related lineups. Officers must now disclose the technology’s limitations and use in arrests, including instances where it fails to identify suspects or produces conflicting results. The revised policies mandate comprehensive training on the software’s risks, emphasizing its higher error rate with people of color. Auditing all cases involving facial recognition technology since 2017 is also required.

The revised Detroit police software policies, enforceable by the federal court for four years, aims to prevent future miscarriages of justice. Phil Mayor, a senior legal expert at the ACLU of Michigan, asserted that Detroit’s revised guidelines would set a benchmark for law enforcement agencies nationwide regarding the ethical use of facial recognition software. Echoing this sentiment, Detroit police officials expressed satisfaction with the policy overhaul. They stated their firm belief that these changes would position them as a national exemplar in facial recognition best practices. The department highlighted the common ground between its mission and that of the ACLU.