Ottawa police admit officers tested Clearview AI facial recognition software – Ottawa Citizen

Ottawa police now say members of the force’s internet child exploitation unit, which investigates child pornography crimes, “created accounts with Clearview AI and tested the efficacy” of the controversial facial recognition software whose unsanctioned use by police forces in this country has prompted internal and external investigations.

City police said Friday that the service was in the process of “polling all of our members” to determine how many officers and in which units had downloaded the software and used it on a free trial basis.

“We expect this will take approximately 4-6 weeks to complete,” police said in a statement. Police could not immediately say in what circumstances the software had been used.

Staff sergeants have been asking members of their units whether any officers downloaded the program. The force has previously said the only sanctioned use of facial recognition software by its officers, and the only one paid for by the public dime, occurred in a three-month pilot of different software last year.

Controversy has swirled around law enforcement’s use of facial recognition technology since a January investigation by the New York Times into a company called Clearview AI. The Times’ story detailed how the company had created a massive database of open-sourced images scraped from websites across the internet, including Facebook. Police could then use the database for comparison with things like surveillance images.

Other facial recognition software, like one piloted by Ottawa police in 2019, compared images not with a database of publicly scraped images, but with internal mugshots that had already been legally obtained by police through arrests and the laying of criminal charges.

This newspaper asked Ottawa police whether they had tested or purchased Clearview AI or any other facial recognition technology on Jan. 24. Nearly three weeks later, Ottawa police responded that they had piloted software, but didn’t immediately name it before revealing the pilot involved a different software. It’s taken five weeks overall for police to preliminarily answer that officers did test Clearview.

The Royal Canadian Mounted Police would first not answer whether it had used Clearview, initially saying it “does not comment on specific investigative tools or techniques. However, we continue to monitor new and evolving technology.”

Canadian police forces began rethinking and reissuing their responses this week, when it was revealed Clearview’s client list had been hacked.

Subsequent reporting by Buzzfeed News and The Toronto Star, whose reporters viewed data on the client list, revealed that Clearview AI’s clients included not just law enforcement agencies abroad and in Canada, but also private companies. That was news to not just the public, but also police services that had not known the extent of Clearview’s use among officers acting without approval from their bosses.

RCMP on Thursday revealed its use of the technology, which prompted the federal privacy commissioner Friday to launch an investigation under the Privacy Act.

Privacy experts across the country have sounded an alarm over use of Clearview AI by law enforcement without a robust legal framework underpinning it.

The Office of the Privacy Commissioner of Canada had already announced it would jointly investigate Clearview AI with privacy regulators in Quebec, British Columbia and Alberta.

Ontario’s privacy commissioner, Brian Beamish, has urged any police service using Clearview AI to immediately halt the practice and to contact the commission.

“We question whether there are any circumstances where it would be acceptable to use Clearview AI,” Beamish said in mid-February, when Toronto police revealed use by its officers.

Ottawa police said Friday that once an internal survey was done and a “full review is conducted on the instances where the Clearview AI was used,” police would report the findings and any recommendations to the police board, which governs the service.

Police have said any plans to permanently implement facial recognition technology would be subject to police board approval and would require the service to fully explore ethical, legal and practical implications.

Police also revealed Friday that, when they did pilot the technology of NeoFace Reveal in 2019, Clearview AI had not been “invited to bid and did not bid in this pilot project.”

Criminal charges did result from use of the approved facial recognition technology pilot in 2019. Police have not said whether charges resulted from the use of Clearview AI.

Chief Peter Sloly, speaking this week, was clear that, while he had no plans to use facial recognition technology, he saw it as the way of the future.

He gave the example of investigating child pornography, the exact work being done by the Ottawa officers who have admitted to trying the software, in which even a simple case can have hundreds of thousands of images as evidence.

Someone has to sit and look at “the most devastating images that a human being can subject themselves to,” Sloly said.

“Why would I take on that human toll when I could use facial recognition and other AI-related technology to do the large data dump, threat assessment, risk assessment, evidentiary assessment and only expose the human being to a limited percentage of that in order to get to a successful investigation.”

The force is expected to answer questions on its use of technology and privacy considerations at a police board committee meeting Monday.

syogaretnam@postmedia.com

twitter.com/shaaminiwhy

ALSO IN THE NEWS

City didn’t set up framework for performance reviews of city manager, auditor general

‘Wanton disregard’ or ‘restrained and deliberate’: Judge to consider closing arguments in Montsion trial

Twitter blows up after Donald Trump mistakenly tags Ottawa Redblacks receiver in tweet