Berlin-based artist Simon Weckert has developed Digital Camouflage, a garment whose pattern is engineered specifically to stop AI object-detection systems from registering the wearer as a person. The work responds directly to a real, current deployment: AI video analysis piloted by Berlin police at Kottbusser Tor. Worn in public, the shirt turns an abstract question about surveillance into something you can put on.
The project’s starting point is Kottbusser Tor itself, where Berlin police began piloting AI video analysis in August 2026 — the first site of its kind in the city. The pattern’s function rests on what’s known as an adversarial attack: a way of manipulating the visual data a machine receives so it fails to interpret what’s in front of it. Earlier anti-detection patches broke down the moment fabric folded or the camera angle shifted, a flaw known as the segment-missing problem — Weckert’s fix is a single continuous texture generated through a method called TC-EGA, printed onto a blend of recycled and standard polyester manufactured in Latvia. Worn on the body, the pattern targets the exact features object-recognition systems look for when scanning a crowd for a human figure, and the algorithm simply stops registering a person.
Weckert demonstrates the effect on video: an open-source object-detection system marks every passer-by in a public square as “person” — bounding box, label, the works — except the person wearing the shirt. The demonstration uses a generic open-source system (YOLO) and makes no claim about how any specific government system would respond — that gap is deliberate, since no one outside those systems can test them. The project’s website also carries a live-feeling map, built from OpenStreetMap’s community-sourced camera data, plotting the surveillance cameras documented around Kottbusser Tor, turning an abstract claim about ambient surveillance into a specific, checkable geography.
A comparable logic — a garment whose real function sits in a layer nobody sees — shows up in a very different context in Axiom Space’s Liquid Cooling and Ventilation Garment for Prada, the inner layer of NASA’s next lunar spacesuit. Both projects treat the body’s relationship to an unseen system — thermal regulation on one hand, machine vision on the other — as the actual design brief, with the visible surface secondary to what the garment is doing underneath.
Eyewear reaches for a related idea from the opposite direction: Oakley’s wrapped-lens prototype with Kiko Kostadinov, an eyewear prototype that covers so much of the face that, seen straight on, it reads as a mask rather than a pair of glasses. Where Weckert’s shirt hides the wearer from a machine, Oakley’s lens hides the wearer from other people — two objects working the same boundary between visibility and concealment from opposite sides.
Weckert is careful about what the piece is not: not a tool for evading police, not a promise of anonymity. He calls it a test image instead — a way of asking how much trust a surveillance system deserves before it’s handed the public square, made concrete by the fact that a piece of fabric can undo it.
What stays with you isn’t the pattern itself but the gap it exposes: a camera can label six strangers “person” in the same frame and miss the seventh standing in plain sight, wearing something that only looks decorative until you know what it’s for.