Public Safety

Bolo Program Shows Canada’s 25 Most Wanted as Life-Sized AI Avatars, and Experts Ask How Accurate They Are

· 6 min read · Public Safety

Bolo Program Shows Canada’s 25 Most Wanted as Life-Sized AI Avatars, and Experts Ask How Accurate They Are

The Bolo Program presented this year’s list of Canada’s 25 most wanted fugitives at a news conference in Toronto on November 19, and the people on it appeared as life-sized, moving AI-generated figures rather than mugshots, CBC News reported. The avatars blink, turn their heads, cross their arms and shrug. Executive director Max Langlois said the aim was to show the fugitives as they might look on the street, at the supermarket or at an airport, adding that if it feels uncomfortable, that is the point.

Bolo, short for “be on the lookout,” has worked with police forces in several of Canada’s largest cities since 2018, using publicity campaigns, new technology and cash rewards to amplify requests for help in finding major crime suspects. This year’s list covers people sought by services in Montreal, Toronto, Winnipeg, Edmonton and Delta, B.C., and the top reward is $250,000. Langlois told CBC that Bolo has used AI images for a few years and builds them from police photos and descriptions, but said the method used to generate them is confidential. The program reports that 13 of its 25 most wanted were arrested in 2025, its best year, and Toronto police say 19 of 28 local cases featured by Bolo led to arrests.

What it means in practice

Bolo Program Shows Canada’s 25 Most Wanted as Life-Sized AI Avatars, and Experts Ask How Accurate They Are — contextual photo

The concern raised by experts is not the goal but the guesswork. David Eliot, a University of Ottawa PhD candidate who studies AI and civil rights, asked how much of each avatar’s build and mannerisms comes from real data and how much from AI or human assumptions, warning that an animation built on thin data could mislead investigators and the public. Tamir Israel of the Canadian Civil Liberties Association compared the avatars to composite sketches, which he said have contributed to wrongful arrests that fall hardest on Indigenous and Black people. On the Bolo website, the AI figures appear first and real photos only further down the page. Israel also asked how many false leads such campaigns generate, which he said does not appear to be tracked.

Both experts argued that oversight rules written for police may not reach a third-party partner, pointing to the privacy commissioner’s 2021 finding against the RCMP’s use of an outside facial recognition tool. The debate joins other questions about AI in Canadian policing, such as Halton police using AI simulations to hire 911 call-takers, and about synthetic likenesses, which Saskatchewan wants to curb with Canada’s first provincial law against deceptive deepfakes. What to watch is whether Bolo discloses how the avatars are made, and whether its police partners set accuracy rules for them.

← Back to AI4Canada