The Quiet AI Surveillance Takeover: Montreal’s Creeping Digital Panopticon
Imagine walking down a Montreal street, unaware that every move is being tracked not by human eyes, but by algorithms parsing your face, your car, even the color of your jacket. This isn’t science fiction—it’s the city’s new reality, where AI-powered surveillance is expanding faster than public oversight can keep up. The recent revelation about Montreal police (SPVM) deploying Genetec’s Cloudrunner cameras isn’t just about a single device; it’s a symptom of a global shift toward algorithmic policing that demands urgent scrutiny.
The Illusion of Transparency
One of the most striking details? SPVM Chief Fady Dagher initially claimed ignorance about the camera’s existence. Let that sink in: the head of the police force didn’t know his own department was installing AI surveillance tools. This isn’t incompetence—it’s structural opacity. When investigative technologies operate in the shadows, accountability becomes a farce. Civil liberties advocate Dominique Peschard’s critique cuts deep: police can share procedural outlines without compromising investigations. The refusal to do so isn’t about safety; it’s about control.
The SPVM’s defense—that they’re only tracking objects like "red coats" or "green bikes"—feels like semantic sleight-of-hand. Even object-based tracking builds databases of behavioral patterns. What happens when "green bikes" morph into "people wearing hoodies near subway stations"? The line between targeted surveillance and mass monitoring is paper-thin, and history shows how easily it blurs.
Safety vs. Privacy: A False Trade-Off
Genetec’s vice-president frames this as a balance between community safety and privacy—a framing I find deeply misleading. This isn’t a zero-sum game. The real question isn’t whether we want safety, but at what cost. The SPVM’s $17,000 investment in Cloudrunner units (with plans to expand) suggests a troubling prioritization: reactive surveillance over proactive community trust-building. Why invest in AI cameras when Montreal’s homicide rate has surged 50% since 2022? Technology can’t solve systemic issues like poverty or gang violence; it merely creates the appearance of action.
What fascinates me most is the cognitive dissonance here. The SPVM claims to respect privacy while storing vast datasets on citizens. They insist there’s no facial recognition—yet their iMotion ROC AI software analyzes video footage for "targeted known elements." Translation: they’re building pattern recognition systems that could easily evolve into de facto biometric tracking. Denial today doesn’t prevent function creep tomorrow.
The Slippery Slope of AI Policing
Montreal isn’t alone. From Edmonton’s facial-recognition bodycams to Clearview AI’s legal battles in BC, North America is sleepwalking into algorithmic authoritarianism. But here’s what most miss: the real danger isn’t malice, but normalization. Each incremental expansion—"just for fighting car theft," "just for finding missing persons"—rewrites social contracts without public consent. It’s the classic "boil the frog" strategy: small changes that lead to seismic shifts in power dynamics.
The Charter of Rights and Freedoms suddenly feels quaint in the face of machine learning. How do we protect against algorithmic bias when AI systems inherit human prejudices? Studies show that even object-based tracking can disproportionately target marginalized communities through proxy variables. A "green bike" in a wealthy neighborhood might mean nothing; in a poorer area, it becomes suspicion.
Toward a Surveillance Society?
Let’s zoom out. This isn’t just about Montreal—it’s about a global reckoning with technology’s role in governance. China’s social credit system feels distant, but the tools being tested in Montreal are Western democracy’s version of soft authoritarianism. The difference isn’t in capability, but in branding.
Personally, I think we’re sleepwalking into a world where privacy becomes a luxury good. Wealthy neighborhoods might fund their own surveillance networks; poorer areas get over-policed. The SPVM’s "voluntary" camera registry proposal—asking citizens to map their home cameras for police access—is the thin edge of this wedge. Participation feels consensual until opting out makes you suspect by default.
The Path Forward (If We Choose It)
What’s the alternative? Radical transparency. Independent oversight boards with tech expertise. Mandatory public audits of all police AI tools. But most importantly, we need to challenge the assumption that more surveillance equals more safety. I’ve yet to see evidence that Cloudrunner cameras reduce violent crime—though they might generate impressive-looking dashboards for bureaucrats.
The deeper issue is cultural: we’ve outsourced our safety imagination to technologists and cops. Instead of asking "How can AI help police?", we should ask "How can communities thrive without needing constant surveillance?" The answer likely involves social workers, not software engineers. Until we confront that, Montreal’s lamp posts will keep quietly sprouting cameras—and our collective freedom will keep quietly shrinking.
