Digital Camouflage is a garment designed to “trick” the computer-vision systems that increasingly watch public space. At first glance it reads as clothing with an abstract, almost hypnotic pattern. In fact, the textile is engineered to confuse the object-recognition algorithms used in surveillance—the class of technology now switching on at Berlin’s Kottbusser Tor, and already humming on streets from London to Los Angeles.
In Berlin, what is being switched on is more than a camera. It is a behaviour scanner: software that sorts what every passer-by does into “normal” and “suspicious”—the first camera legally allowed to film the city’s public space without any specific cause. To isolate a punch, sister systems have to classify dancing, hugging and countless other harmless gestures too; everyone is read, all the time, under a quiet reversal of the presumption of innocence. And Kottbusser Tor is only the template: three more locations across the city are next.
Berlin is the beginning of a curve whose end is already visible elsewhere. In the United States, automated surveillance arrived years ago: well over a hundred thousand networked cameras log vehicles and movements across thousands of towns, their data flowing into databases that police—and federal agencies—can search without a warrant. And 2026 has become the year of the pushback: dozens of communities have voted to cancel their camera contracts, civil-rights organizations are campaigning to take the networks down, and the question of who is allowed to read the public street has entered the mainstream. This work stands in that moment, on both sides of the Atlantic.
The pattern of Digital Camouflage is built on what is known as an adversarial attack: a way of manipulating the visual data a machine receives so that it fails to interpret what is in front of it. Worn on the body, it targets exactly the features these systems look for when searching for a human figure. The algorithm no longer registers a person. The wearer becomes, in effect, invisible to the machine—present to every human eye, absent to the model. Whether any specific government system reacts the same way, no one outside can verify—and that opacity is precisely part of what the work makes visible: cities are deploying systems that cannot be independently tested, while the same class of technology is undone by a piece of fabric. The work was documented on site at Kottbusser Tor, in photo and video.
What does a city get in return? Decades of research suggest that cameras rarely prevent crime—they displace it into the next street, or produce a feeling of safety rather than safety itself. A camera cannot step in when something happens; worse, it teaches bystanders not to step in either—someone official is watching, so no one has to. Meanwhile, the manufacturer of Berlin’s system is being kept secret.
Digital Camouflage is not a tool for evading the police, and it is not a promise of anonymity. No pattern can promise to defeat every system under every condition—but worn in public, it is a visible signal that being read by machines is not something everyone consents to. A watched square changes people before any alarm ever fires. When systems that cannot be publicly tested can be undone by a piece of fabric, how much trust should we place in them before we hand them the public square—in Berlin, or anywhere else?
A share of the proceeds from every shirt goes to organizations defending digital civil rights and fighting the expansion of surveillance.
The reference system was YOLO, one of the most widely used open-source real-time object detectors — the same class of neural network deployed in video surveillance. The pattern was developed in an adversarial loop against it: generate a candidate, show it to the detector on a person, measure how confidently the system still finds a “person”, adjust, repeat — until detection confidence collapses below threshold.
Project Page | Simon Weckert | Instagram
