An Italian company, Cap_able, has developed a unique solution to evade facial recognition technology. They sell knitted garments with patterns designed using AI, which confuses the recognition system and causes it to identify the wearer as an animal, instead of a human.

Facial recognition technology has made great strides in recent years, but it is not without its weaknesses. Instead of using masks or makeup, one can now wear a Cap_able sweater to evade detection. The YOLO algorithm, which is commonly used in facial recognition, is unable to recognize the wearer if they are wearing the Cap_able sweater and instead, surrounds them with a frame.

A time-limited effectiveness

Rachele Didero, co-founder of Cap_able, developed the unique patterns used in the company’s garments as part of her doctorate at the Politecnico di Milano. The goal was to deceive the most commonly used algorithm in facial recognition, YOLO. The idea is simple: before recognizing faces, the AI first identifies humans in the image. By wearing these sweaters with specific patterns, the algorithm instead identifies the wearer as an animal, such as a dog, zebra, or giraffe, and therefore does not classify the face as human. The patterns are generated with the help of AI and have a success rate between 60-90%.

However, the effectiveness of these garments may be limited. YOLO is constantly updated, and could potentially recognize people wearing the garments overnight. Cap_able is working on improving the patterns to better confuse the AI, but recognizes that this is a real arms race and that the clothes will have limited effectiveness over time. The garments are also expensive, with knitted pants costing 277.50 euros and long-sleeved sweaters costing 390 euros.