On Thursday night, a drone hovered over Canada’s most secure prison. The place was the Special Handling Unit in Sainte-Anne-des-Plaines outside Montreal, the country’s only “supermax,” built in 1984 for people who cannot safely be held even in maximum-security prisons: at various times it housed serial killer Clifford Olson and Hells Angels boss Maurice “Mom” Boucher. The drone dropped a McDonald’s box by a cell window — nuggets, fries, sauce — and a bottle of alcohol. Neither Correctional Service radar nor surveillance cameras noticed it; officers themselves reported the delivery after finding the inmate eating in the morning. According to union head Mike Bolduc, the prisoner “was laughing in the faces of the guards — like, I can bring anything I want in here.”
The story spread through the channels like a joke: a prison where, it seemed, not even a bird could get through was accepting McDonald’s deliveries. But the funny image conceals a far less funny mechanism: for its eighth year, Canada has been building air defense over its penitentiaries, spending millions, legalizing specialized jammers in 2025 and buying radars — and in 2026 drones are still delivering drugs “like mail straight to the window.” Here’s why a little plastic quadcopter worth a couple thousand dollars still outplays state air defense.
Radar is a grown-up, powerful thing. But it has an Achilles’ heel: it sees well what it has been prepared to see. The DSIAC report (an analytical bureau of the U.S. Department of Defense) on countering small drones puts it plainly: the same radars that reliably track large aircraft “miss” small, slow Class 1–2 systems. There are three reasons, and all three favor the smuggler.
First: radar cross-section, RCS. A DJI Phantom 4, according to DSIAC estimates, has an RCS of around 0.02 m² — meaning the drone reflects so little energy that against hangars, fences, trees, and ground, it is almost indistinguishable from noise. Second: slow speed. Classic radar filters screen out “clutter” targets such as birds, and a quadcopter hovering over a yard fits that filter perfectly. Third: terrain. Amid buildings, the radar background is dirty, while modern drones generally fly low, using those very structures as cover.
There is a fourth reason, far less technological. After the Sainte-Anne-des-Plaines incident, the UCCO-SACC-CSN correctional officers’ union pointed out that one or more of the facility’s surveillance cameras were simply broken. And in September at Stony Mountain, Manitoba’s only federal prison, union president Jake Suelzle sneered: “CSC just invested serious money in a drone detector — and the detector demonstrated its value by detecting absolutely nothing.” The alarm did not go off; drug packages worth $1.8 million were spotted by officers’ eyes.
In short, the blind spot is not even a single technology but a chain: radar tuned for major threats, cameras, some of which do not work, and people, of whom prisons always have fewer than they have perimeter.
The Canadian McDonald’s episode looks eccentric, but the numbers show it is not a malfunction but the norm. Quebec has published reports on drone events in its prisons since fiscal year 2021/22: then there were 695 drone activities and 409 contraband deliveries. In 2025/26 — 1,521 events and 922 deliveries. Not only is the raw number rising; the effectiveness of the defenses is falling too: in 2022/23, 83% of packages were intercepted; in the first quarter of 2026/27, 68%.
The federal Correctional Service of Canada (CSC) reports publicly less often, but the Toronto Star, citing 2025 data, wrote of 1,064 drone episodes in federal penitentiaries in an incomplete year, against 899 the year before. For scale: England and Wales counted 1,712 drone incidents from April 2024 to March 2025, up 43% year on year; Georgia, U.S. — nearly 400 incidents in 2025, and the state’s Department of Corrections seized aircraft capable of lifting loads of 100 kilograms.
And this is not “toy delivery.” In February 2026, near Winnipeg, the RCMP caught a driver whose passenger was waiting for a drone package: methamphetamine and cocaine worth $751,000, plus a second package worth $600,000. On October 9, 2026, federal Millhaven seized a package worth $739,950: tobacco, “shatter” (cannabis concentrate), crystal meth, knives, phones — again, a suspected drone drop. The same week as the McDonald’s delivery. The stream is continuous, and the nuggets are simply its most photogenic part.
Why would anyone be willing to fly a drone over a prison at night at all? The answer is margin. In the prison economy, an iPhone bought for $1,000 is resold for up to $6,000; in Latin American prisons (an InSight Crime survey in El Salvador), a phone that costs $20 outside goes for $200 inside, while a gram of cocaine that costs $25 goes for $250. A cheap quadcopter that costs a couple thousand dollars and brings in goods worth hundreds of thousands per run pays for itself on its first flight; even one successful delivery out of ten covers the risk to the technician sitting in a car a kilometer from the fence.
The delivery operation itself has long been industrialized. At HMP Wandsworth in London, police seized a £6,000 drone with a flight time of 40 minutes and the capacity to carry four loads per run. At England’s HMP Garth, night drone runs became so regular that one inmate compared the prison to an airport. The Guardian, citing surveys, added one expensive detail: gangs headhunt civilian drone pilots — people who may never have broken the law, but know how to fly a GPS route in the dark.
The British answer is nets stretched over prison yards — £40 million was invested in them in 2025 alongside window reinforcement — and the same option is being proposed by the Canadian union. Physically, it works: a drone that cannot descend to a window is useless. But every yard needs a net, and there are many prisons, while the question of budgetary will is the same one that has killed many other counter-drone programs.
Canada’s bureaucratic timeline is a story of its own — about the speed of the state versus the speed of the market.
2017: CSC intercepts a drone package worth $26,500 at Matsqui Institution in British Columbia — and realizes that the sky above prisons is no longer a hypothetical. October 2018: the ISED agency opens the “Preventing Contraband Delivery via Air and Ground” competition. July 2019: CSC allocates $6 million and, together with the National Research Council, tests 60+ radar detection systems at six sites — including, incidentally, Stony Mountain.
Then silence: the pilot project was supposed to conclude by March 2023 with a report and recommendations. March 2025: the government finally registers SOR/2025-63 — an order legalizing “jammers” for the first time (electronic warfare devices that jam drone control frequencies) for CSC prison staff and Quebec’s correctional system. September 2026: at Stony Mountain — the very site that was in the pilot project! — a new detector fails to see a drone. October 2026: SHU, nuggets right under the radar’s nose.
All told, nine years passed from the first recognized incident to a functioning legal tool. In that time, the market went from homemade camera-equipped quadcopters to machines carrying a hundred kilograms. To say the state is one step behind is to flatter it.
Every iteration of this confrontation has a ceiling, and it is not technological. According to DSIAC, Ku/Ka-band radars with AESA arrays can see small drones at 1–3 km; micro-Doppler processing distinguishes propellers from birds; RF detectors catch the control signal. The technologies exist and cost reasonable money. The jammers legalized in Canada are physically capable of bringing a drone down. The problem is that all of it is a patch over a fundamental asymmetry: the defender must cover everything, always; the attacker need only succeed once.
The asymmetry is reflected in prices as well. A detector able to tell a bird from a copter costs three orders of magnitude more than the copter. Every new measure — nets, jammers, GPS spoofing — raises the cost of delivery but almost never makes it impossible: the drone flies higher, drops its cargo by parachute, uses autopilot along a prerecorded route with no radio link to the pilot at all — and there is nothing left to jam; all that remains is to search.
The irony of this story is that its one character who did not lie in a single detail is the SHU inmate. He really can bring anything in here. One question stands in place of a happy ending: when will “anything” stop being nuggets?