After an incident involving a misidentification of a 32-year-old Black woman, Porcha Woodruff, who was wrongly charged with robbery and carjacking, the Detroit police chief has announced new policies governing the use of facial-recognition technology. Chief James White indicated that the case was a product of facial-recognition-derived leads combined with “very poor” police work.
The arrest incident involved Woodruff, who was eight months pregnant at the time and was apprehended while preparing children for school. She was linked to a January robbery and carjacking via facial-recognition technology. Although Woodruff denied involvement, charges were later dropped due to the victim’s absence in court.
Chief White firmly stated that officers would no longer be permitted “to use facial-recognition-derived images in a photographic lineup. Period.” Additional policy changes include a mandate for two captains to scrutinize arrest warrants when facial recognition is applied. These new guidelines will soon be introduced to the Detroit Police Board of Commissioners.
According to ABC News Reports, White emphasized the need for evidence beyond just the technology when suspecting someone of committing a crime, noting the technology’s limitations. Reflecting on the incident with Woodruff, he said, “We recognize we have to do better, and there will be accountability on this mistake.”
Separate from this case, research findings from Scientific American have shown concerns about the potential racial biases of facial recognition technology. The study indicates that law enforcement relying on such technology tend to disproportionately arrest Black individuals. Several factors, such as underrepresentation of Black faces in training data sets and officers’ existing biases, have been identified as contributors.
To address concerns about the misuse of facial recognition and other AI technologies, several U.S. cities, including San Francisco and Boston, have imposed restrictions or outright bans on its use by government entities. At a national level, the Biden administration proposed the “Blueprint for an AI Bill of Rights” in 2022. While setting forth principles, the blueprint is nonbinding. Meanwhile, congressional Democrats have reintroduced the Facial Recognition and Biometric Technology Moratorium Act, which seeks a temporary halt to law enforcement’s use of facial recognition until proper regulations can be established.
Recent research underscores the importance of diversity in the design and development of facial recognition systems. Current software seems better at identifying individuals from the same race as the software developers, most of whom are White males in the U.S. This phenomenon is attributed to an “own-race bias” unintentionally incorporated into algorithms.