Police Chief Craig and Mayor Duggan have assured the people of Detroit that the Detroit Police Department won’t use facial recognition technology to track Detroiters and visitors in real time. They have also assured us that there would be strict guidelines for the use of this technology with still images.
While this may seem comforting, we at DRACO retain serious concerns regarding any continued use of this technology by law enforcement.
Credibility is central to any policy statement or promise. The process of establishing a policy on this matter just became an agenda item for the Board of Police Commissioners a few months ago, but the Office of Contracting and Procurement has been looking into the purchase of real-time and still facial recognition systems since 2015. Systems with this capability were already being deployed in 2017, and the Detroit Police Department apparently withheld details of the facial recognition component of Project Green Light from the BOPC and Green Light partners when it was presented to them and the public.
The significance of this is evident in that some of the strongest supporters of project Green Light are now vocally opposed to facial recognition technology. Clearly, withholding this information served to enable the broad deployment of a system that would be compatible with the full-spectrum of facial recognition applications without informed oversight.
Policies assuring limitations on use, or the enumeration of extreme exceptions, is inadequate when the capability to do otherwise is ever-present. Promises of compliance ring disingenuous when past transparency is so lacking. Add to this that the oversight organization for the DPD has steadfastly exercised gross indifference for clearly stated laws like the Open Meetings Act.
Even a best case scenario, where policies are earnestly complied with, leaves open room for a complete reversal amidst a crisis or a change in leadership. Furthermore there is the likelihood of abuse by staff member entrusted with this technology. Such abuse by colleagues was what motivated Edward Snowden to blow the whistle on the NSA, for instance.
But even if all persons are presently serving with the utmost integrity, technology that enables invasions of privacy can easily be turned against the populace when there is a regime change or shift in power. Just turn to the headlines to see how the relatively free people of Hong Kong now see themselves on the brink of being absorbed into the malevolent surveillance state that is mainland China.
But what about those still images? The well-documented failure of this software to correctly identify people of color is just as severe with still images as with real-time systems. This problem has earned the use of facial recognition technology the label “Techno-racism.” Clearly there is a serious disconnect between some decision makers and members of the community who wonder why a system that chronically misidentifies African Americans is being deployed in a predominantly African American city.
Misidentification can lead to hardship for the innocent. Chief Craig has assured us that a human being will make the final decision on identification, but there is bound to be a bias in favor of agreeing with the initial impression that could reinforce misidentification in borderline cases. The counter argument is that these are fixable problems, but the simple fact is that the problem is not fixed, and the DPD should suspend use of facial recognition technology on still images at least until this problem is resolved to the satisfaction of experts and the majority of voters in Detroit.
Therefore, it is the position of DRACO that there be a moratorium on the deployment of facial-recognition-ready hardware and the installment of facial recognition software from DPD computers. Its use should be discontinued for the foreseeable future.
This information was read at a forum at Red Door Digital in Detroit which DRACO supported:
Detroit Residents Advancing Civilian Oversight 