Montreal grapples with privacy concerns as more Canadian police forces use facial recognition

From shopping malls to smartphones to airports, facial recognition technology is becoming an increasingly common part of daily life.

But should the emerging technology be used by law enforcement?

One city councillor wants Montreal to follow the lead of San Francisco, which voted to ban the use of facial recognition software by the city and police earlier this year.

Independent Coun. Marvin Rotrand told CBC Montreal's Daybreak that the city's bylaws should be updated so that they can keep pace with advances in artificial intelligence, to protect citizens' "privacy and the protection of our democratic values."

He's preparing a motion that would put a moratorium on using the technology until clear rules are put in place. It will be debated at the city council meeting on August 19.

The technology uses banks of images, such as mugshots and government-issued IDs, to map a geometric pattern on an individual's face that can be used to compare against other images such as surveillance footage.

Using this pattern, faces seen from different lighting and angles can be compared to those in their image bank using artificial intelligence.

Facial recognition being used by Canadian police

Montreal Mayor Valérie Plante says council will debate the use of such technology by the city and its police force.

"We do want to have this discussion," Plante told reporters on Wednesday.

"We need to be cautious how we use data."

Montreal police would not say whether or not it is already using facial recognition technology, but it is being used by police elsewhere in Canada.

The Toronto Police Service said last year a pilot project using the technology was an "immediate success" in helping identify suspects.

The force spent $451,718 to buy their system, using funding from a policing grant from the Ontario government.

Calgary police began using the technology in 2014.

South of the border, Anne Arundel County Police said the technology was used to confirm the identity of the man who perpetrated the June 2018 mass shooting at the Capital Gazette newspaper in Annapolis, Md.

However, privacy advocates argue that innocent people can get swept up as verifying someone's identity gets outsourced to AI.

"Once an arrestee's photo has been taken, the mugshot will live on in one or more databases to be scanned every time the police do another criminal search," writes the Electronic Frontier Foundation (EFF), a San Francisco-based digital rights advocacy group.

The group argues this means that people found not guilty in court will still have their face in the system, and that the larger the bank of faces, the more risk there is of returning a false positive.

Is it reliable?

There is also criticism that facial recognition technology is more accurate matching white faces than black ones.

"The algorithms are typically created by white males and they tend to do a much worse job with people who aren't of the same bias as the algorithm maker," René Ritchie, editor of technology blog iMore, told CBC Montreal's Daybreak.

A 2012 study by the Institute of Electrical and Electronics Engineers concluded that more work needs to be done to shrink that demographic gap in the reliability of the technology.

Montreal has become a hub for artificial intelligence research, including by Microsoft and Google.

Idemia, which according to the EFF is "one of the largest vendors of face recognition and other biometric identification technology in the United States," also has a presence in Montreal.

Rotrand wants the federal government to create laws governing the use of facial recognition, but for the time being he says it's important that city council send a clear message to police that they aren't free to use it as they please.

Even as the technology becomes more reliable, he says the threat to Montrealers' privacy will remain.

Paired with security cameras, "the police would now have a tool that would basically follow you all day long," he said.