How Simon Weckert’s "Digital Camouflage" Garments Use Adversarial Attacks to Fool AI Cameras
The Berlin-based designer creates a physical countermeasure to automated tracking, utilizing generative visual noise to make wearers invisible to AI person detectors.
Summary
Simon Weckert debuted "Digital Camouflage," an apparel collection engineered to trick AI surveillance.
Generative patterns prevent camera detection from any angle or fabric fold.
Digital printing applies visual noise onto recycled polyester textiles.
Berlin-based artist Simon Weckert is pushing back against the rise of ubiquitous digital surveillance with his 2025 installation. The designer officially unveiled “Digital Camouflage,” a conceptual garment collection that serves as a wearable shield against AI-driven tracking systems. As artificial intelligence continues to dominate public spaces through automated tracking cameras, personal privacy has become a highly debated topic. This project explores the boundaries of anonymity and visibility in modern society. By merging computational noise with everyday apparel, Weckert offers a physical countermeasure to digital overreach.
At first glance, the clothing appears to feature nothing more than a striking abstract graphic pattern. The collection is actually a highly engineered tool specifically developed to disrupt object-recognition algorithms used in security and autonomous systems. This design approach relies on the concept of an adversarial attack. The technique deliberately manipulates the visual data received by a machine so that it incorrectly interprets the scene. When an individual wears the collection, their body essentially becomes invisible to AI person detectors because the software completely fails to identify the human figure amidst the high-frequency visual noise.
Previous attempts to evade machine vision typically relied on fixed printed patches or isolated graphics. Those early designs shared a critical vulnerability known as the segment-missing problem, where the illusion breaks down the moment the fabric folds or the camera angle changes. Weckert solves this technical hurdle by deploying a seamless Adversarial Texture that covers the entire surface of each piece. A specialized generative AI method called TC-EGA generates the continuous pattern. This computational approach optimizes a tileable textile design that constantly produces false visual features regardless of the viewing perspective.
Physical production of the garments matches the meticulous nature of the software engineering behind them. Manufacturers in Latvia construct the collection using a durable blend of 65 percent recycled polyester and 35 percent standard polyester. Developers apply the computational patterns through precision digital textile printing to ensure the visual noise remains intact across the fabric. These resulting garments effectively blend digital pattern generation with tactile textile production to challenge modern recognition systems.
