Ottawa used facial recognition on ~2.95M travelers at Toronto's Pearson airport between July and Dec. 2016, experts say with no public discussion in advance
In an effort to identify potential deportees, the federal government quietly tested facial recognition technology on millions … Tweets: @lex_is and @moore_oliver Tweets: Lex Gill / @lex_is : Extraordinary new reporting out today from @tom_cardoso on the use of facial recognition at Canada's largest airport. https://twitter.com/... Oliver Moore / @moore_oliver : If you passed through international arrivals at Toronto's main airport, the country's biggest, in the second half of 2016 you were unwittingly part of a huge facial recognition sweep. Exclusive by our @tom_cardoso and @Colinfreeze https://www.theglobeandmail.com/ ...
Context & Ripple Effects
Canada's border agency was running the same play as its US counterpart, just quieter: years before American coverage documented how the government and airlines began scanning faces of people not suspected of any crime, Ottawa tested facial recognition on roughly 2.95 million international arrivals at Toronto Pearson between July and December 2016 to identify potential deportees — with no public discussion in advance. The public learned of it only now, through Globe and Mail reporting, five years after the fact.
The timing matters because it lands amid Canada's push to brand itself an AI leader — the country has unveiled a national AI strategy touting 250,000 jobs by 2031 and a CA$500M fund for homegrown startups — while its own government's most consequential biometric deployment was disclosed by journalists rather than by any oversight process.
First-order effects
- Every traveler who passed through Pearson's international arrivals in the second half of 2016 was enrolled in a facial recognition sweep without notice or consent, and Ottawa now faces accountability questions over a program it deliberately kept undisclosed.
- Privacy advocates such as Lex Gill are pressing the disclosure angle publicly, turning a five-year-old test into a live political liability for the federal government.
Second-order effects
- The US record undercuts the efficacy argument Ottawa would otherwise lean on: CBP's own data showed 23M-plus scans in 2020 caught zero imposters at airports, so Canada's deportation-focused test invites the same question about what the scanning actually achieved.
- US agencies have since moved to consent-framed deployments — the TSA markets its pilots as voluntary and is scaling toward ~430 airports — which puts quiet, no-consent tests like Pearson's on the wrong side of an emerging norm that critics will cite (TSA's expansion claims 97% accuracy across demographics).
Third-order effects
- If the pattern holds, biometric screening of non-suspects becomes normalized by deployment-first policy on both sides of the border, with legal and consent frameworks arriving years later and only after journalistic exposure forces the issue.
- Governments promoting themselves as AI leaders face a widening gap between their innovation strategies and their governance practice — the Pearson test suggests public-safety AI will keep outrunning disclosure unless oversight bodies, not reporters, become the ones who reveal these programs.
The trend: Governments are deploying facial recognition on mass traveler populations first and disclosing it years later, while consent-based framing and accuracy claims emerge only after public exposure forces the conversation.