Picking out a face in the crowd: Toronto police considering facial recognition technology
Privacy advocates warn that the technology raises big questions about surveillance
The Toronto Police Service says it is exploring the use of facial recognition technology as a tool to help solve crimes.
It's part of a dramatic shift in law enforcement techniques that could make law enforcement more efficient by cutting down on the time it takes to identify suspects. The software has already been adopted by Calgary police, and is widely used by law enforcement agencies in the United States as a means of matching crime scene images to mug shots.
But for all its abilities, privacy advocates caution that the technology raises big questions about surveillance, and has potential implications for members of the public who aren’t suspects of a crime.
So far, Calgary is the only police department in the country to fully implement facial recognition software, though the company that supplied them says other police in Canada are interested.
In a press conference last week, Insp. Rosemary Hawkins said Calgary police won't use the technology to identify law-abiding members of the public. Instead, she advocated its crime-fighting advantages — mainly, that the previously gruelling task of identifying criminals by searching through binders of mug shots can now be done by the click of a mouse.
"With the implementation of this facial recognition software, we will now be able to do this within minutes and seconds," she said.
Indeed, in the United States, some of the results have been remarkable. A few months ago, authorities in Florida used it to catch convicted murderer James Jones, who had escaped from prison in 1977 and had been living under a different name ever since. Officials used facial recognition software to match old military pictures of Jones to his driver's licence, registered under his fake name.
In cases like these, the technology has clear advantages, says privacy expert Christopher Parsons, a fellow at the Munk School of Global Affairs at the University of Toronto.
"Serious crimes — rapes, murders, manslaughter — these are the kinds of crimes that must be brought to justice," he says. "But for other crimes, lesser crimes, maybe those aren't the situations where we [should] use these really efficient, high-tech systems." The risk, he says, is that "it starts … criminalizing a large portion of the population."
Toronto police haven't yet indicated how or when they might want to use the technology. Last winter, the police service called for consultations, seeking information about the "usefulness of implementing a facial recognition system within the TPS environment." Calgary police have said they will use their new facial recognition software in a wide range of cases, from homicides to robberies.
The risk is when facial recognition "starts … criminalizing a large portion of the population."
Police aren't the only organizations to employ this type of technology. Some department stores and retail chains also use it to catch repeat shoplifters. But Parsons points out there is a difference between private individuals capturing images and the police.
"[Private individuals] don't have the power to arrest," he says.
Not long after Calgary police made the announcement about the technology, the privacy commissioner of Alberta launched an investigation, seeking information on how the service plans to use, store and safeguard the information they gather.
Even in the U.S., where privacy laws tend to be more relaxed than in Canada, the courts are catching up with facial recognition technology. Just this month, a judge ruled that the facial recognition database held by the Federal Bureau of Investigation deserves transparency oversight.
There is evidence that privacy advocates and government agencies can work together to avoid these potential breaches of civil liberty. Former Ontario privacy commissioner Ann Cavoukian worked with the Ontario Lottery and Gaming Corporation (OLG) to develop a privacy-friendly facial recognition system for casinos.
There, the technology was put in place for gambling addicts, who wish to exclude themselves from casinos by allowing security cameras to recognize them if they try to enter.
"It's positive consent from a privacy perspective," says Cavoukian, who also asked that the data be encrypted so it couldn’t be used by anyone outside of the OLG who might somehow obtain it.
As for all the faces of people who haven't agreed to the program, Cavoukian asked that they not be stored in the program. "They agreed to that," she says.
From this experience, Cavoukian believes it is possible for police to follow similar privacy guidelines when adopting their own software.
"What I would ask for [from] law enforcement contemplating doing this in Canada, is [for them to] step up to a higher standard," she says. That includes considering some basic questions about the storage and purpose of the data they gather, such as, "How long are you going to retain this information in your database? Are you going to use it for other purposes?" she asks.
Cavoukian says: "If in Canada these systems are just being developed now, it's an ideal time to ask these kind of questions and try to ensure that we can safeguard the privacy of law-abiding citizens, while identifying the potential terrorist."