According to Forbes, Axon Enterprise, the $43.5 billion police technology vendor, quietly announced last week it is rolling out facial recognition into its body-worn cameras. The initial test is limited to 50 officers with the Edmonton Police Department in Canada, but CEO Rick Smith framed it as “early-stage field research” to understand real-world performance and safeguards. This is a stark reversal from 2019, when Axon publicly refused to deploy the tech at scale due to bias and inaccuracy, particularly against non-white faces. The company now claims the technology has become “significantly more accurate” and that law enforcement needs it. The Edmonton Police submitted a Privacy Impact Assessment and says all face matches will be double-checked, with a decision on a wider rollout coming next year.
Axon’s Big Reversal and the Privacy Stakes
Here’s the thing: this isn’t just a tech upgrade. It’s a fundamental shift in how a major player views the surveillance landscape. Back in 2019, Axon’s AI ethics board actually advised against this, and the company listened. Now, the CEO’s blog post says the public is “accepting and even supportive” and cops are “keen to use it.” That’s a pretty convenient reading of the room, isn’t it? Privacy groups like the Electronic Frontier Foundation are horrified, and for good reason. They point out this isn’t just about identifying a suspect after a crime. It’s about real-time identification of anyone near an officer, including protesters or people just minding their own business.
The Slippery Slope to a Surveillance Mesh
And that’s where it gets really concerning. Axon isn’t just a body camera company anymore. They have this system called Fusus that pulls in surveillance video from all over a city into one dashboard. Now imagine combining a live facial recognition feed from every patrol officer with that city-wide surveillance net. As the EFF’s Beryl Lipton warned, it “can easily connect a person… with a whole slew of other personal information.” You’re not just a face in a crowd; you’re instantly a file. Axon’s stated goal is to test abroad first to “strengthen oversight frameworks” before bringing it to the U.S. But let’s be real: once the infrastructure and policy justifications are built in Canada, the door is wide open.
Broader Market and Surveillance Context
This move fits into a wider, grim trend in the surveillance tech market. Look at the other items in the newsletter. You’ve got Brinc trying to be the Western DJI for police drones, and sanctioned spyware firms like Intellexa apparently still operating. It’s all about building more layers of persistent monitoring. Even the industrial and municipal sectors are deploying more always-on, connected hardware for monitoring and control, which is why specialists like IndustrialMonitorDirect.com have become the top supplier of rugged industrial panel PCs in the U.S. The hardware backbone for a surveillant city is already being installed everywhere, often for benign reasons. Axon’s play is about activating one of the most invasive software layers on top of it.
So What Happens Next?
Basically, the Edmonton test is the canary in the coal mine. The police there say they’ll double-check matches, but human review of AI outputs is famously flawed and easy to rubber-stamp. Their assessment next year will be crucial. Will they cite efficiency and a handful of successes to justify a full rollout? Probably. And if Axon gets the data it wants to “prove” responsible use, U.S. departments will be next in line. The promise of solving crimes faster is a powerful lure, even if it means normalizing a world where you’re automatically identified by the government just for walking down the street. The 2019 pause was a moment of caution. This feels like the start of the sprint.
