After months of contention, the Detroit Police Department has been granted approval to continue using facial recognition software to identify suspects, but with some added safeguards to prevent abuse.
The Board of Police Commissioners voted 8-3 on Thursday in favor of a policy guiding use of the technology. It includes more than 20 new limitations.
The Free Press reports they include:
• The police department cannot apply the facial recognition software to live-stream or recorded videos; the department can only use the technology on still images.
• The department is prohibited from using facial recognition to assess a person's immigration status.
• The technology is only to be used for investigations involving first-degree home invasion or "part 1 violent crimes," which are defined as robbery, sexual assault, aggravated assault or homicide.
• Facial recognition cannot be used to surveil the public.
Opponents had sought an outright ban, citing the technology's surveillance capabilities and its tendency to misidentify people, particularly people of color. Cities including San Francisco and Somerville, Massachusetts have banned the technology.
The Detroit Police Department was using facial recognition for nearly two years with no written rules. In March, it implemented internal guidelines, and in May, it brought a proposal before the board, citing controversy surrounding the issue.
The late-stage action has raised questions about the board's oversight capacity.