The purpose of the state’s facial-recognition system is sound. It is designed to tap data, or photographs, from the Ohio Bureau of Motor Vehicles to help law enforcement authorities prevent and solve crimes. It also carries the potential for abuse, a concern that resonated recently when the Washington Post reported that the FBI and Immigration and Customs Enforcement have used state driver’s license databases to look through millions of photos of Americans without their knowledge and consent.
Ohio is one of those states. Thus, Dave Yost, the state attorney general, responded appropriately in launching a review of the system here, seeking to bring helpful clarity and transparency.
On Wednesday, the attorney general announced his findings. They are reassuring, the review concluding “there is no evidence of federal misuse … such as for mass surveillance, broad dragnets or other illegitimate uses.” Those tapping the system, in operation since 2013 and consisting of 2011 driver’s license photographs, mostly are local and state authorities.
The review examined usage from January 2017 to July 2019, totaling 11,070 searches. Among the heaviest users were the Hamilton County sheriff’s office (4,180 searches), the state homeland security office (1,984), the Cincinnati Police Department (896), the Bureau of Motor Vehicles (862) and the Columbus Police Department (768). The Akron Police Department conducted 185 searches.
Federal usage amounted to 4 percent of the total, the border patrol with 142 searches and ICE with 87.
Yost reported “everything is being done well.” That follows, considering the guidelines set by Mike DeWine, now governor, who was attorney general at the inception. DeWine formed an advisory committee, which recommended improvements.
At the same time, Yost had good reason to add: “I want it to be done better.” He has proposed that the 4,549 people with access to the system receive training. Until then, their requests will go through the state Bureau of Criminal Investigation, as federal inquiries currently do. The attorney general will put together a task force of law enforcement officers, technology experts and privacy advocates to aid in developing the training. The panel also will advise on rules for operating the system.
Authorities typically access the facial-recognition system when they have a photograph of a suspect, but they do not know the identity. In that way, the attorney general is correct in saying the system is a “starting place.” He sees the training as key in dealing with the limitations of the system.
Those limits have come into view through recent studies. In January, research at the Massachusetts Institute of Technology found that a facial-recognition program from Amazon incorrectly identified darker skinned women as men nearly one-third of the time. Another study, conducted by the Florida Institute of Technology and the University of Notre Dame, reported a higher rate of false matches among African Americans than for whites.
The study noted that in altering the software to achieve a more accurate read for blacks, the percentage of false matches for whites climbed. Researchers caution that skin color may not be the decisive factor. The problem could involve such things as facial structure or hairstyle.
The state facial-recognition program is one of 22 components of the online Ohio Law Enforcement Gateway, or OHLEG, allowing authorities to share information to advance investigations. What the attorney general’s review reinforces is that facial recognition is a work in progress. So it matters that the message is clear to law enforcement organizations: The system isn’t a substitute for other police work. Any abuse brings the grave risk of eroding public confidence.
Here is an investigative tool that requires care, or what the attorney general proposes in moving forward.