Rahul Somvanshi
—
A growing number of fashion designers are building clothes that are specifically engineered to confuse artificial intelligence surveillance cameras. The garments are real, the technology behind them is real, and the civil liberties questions they raise are getting louder.
The idea is called adversarial fashion. The clothing uses carefully designed patterns — specific shapes, colors, and repeated visual arrangements — that exploit known weaknesses in the computer vision systems used by automated cameras. These systems, including widely used AI models like YOLO and Faster R-CNN, process images by analyzing patterns of pixels. When you introduce the right kind of visual noise into that frame, the system either misidentifies the person in it or fails to detect them at all. The camera might register the wearer as an animal, a piece of furniture, or nothing at all.
Nick Tidball, co-founder of a company called Vollebak, believes this category of clothing is approaching mainstream consumer interest. He argues that “anti-surveillance feelings are so widespread that all it would take is for a single celebrity to wear one of these garments… to a high-profile event for it to take off.” Daniel Preuß, co-founder of Urban Privacy, says his brand uses large-scale prints, asymmetrical cuts, and streetwear-shaped designs to confuse the algorithms. The company’s flagship product, called the “Urban Ghost” coat, has LEDs stitched into the hood that emit infrared light — the kind of light that night-vision surveillance cameras pick up and “freak out” from, because the system sees multiple apparent faces at once.
Cap_able, a company founded by Rachele Didero in 2018, makes knitted garments with adversarial patterns printed directly into the fabric. When Didero started, she says people assumed she was “designing masks to rob banks.” In the years since, the category has moved well beyond novelty. She says interest has “rocketed in the last few years” as younger consumers become more conscious of AI tracking in public spaces.
A 2026 UK poll found that almost 60% of respondents viewed facial recognition technology as “another step towards turning the UK into a surveillance society.” This sentiment has been reinforced by documented cases showing that facial recognition systems misidentify people from Black and Asian backgrounds at higher rates than white people.
But the technology has limits that honest coverage should acknowledge. Dr. Jennifer Bell, a senior lecturer in creative AI and fashion at Nottingham School of Art & Design, is blunt about it: “None of these products are tried and tested, and a lot of these surveillance technologies can deal with a little resistance.” She adds, though, that “even if the designs don’t necessarily work perfectly, fashion is also a visible sign of resistance.”
There’s also a political risk Tidball raises that most coverage misses. If adversarial clothing ever became genuinely effective at defeating surveillance at scale, it “could get political very quickly” — potentially prompting governments to regulate or outright ban it.
Whether wearing such clothing is legally permissible depends on the jurisdiction and on whether the garment merely uses visual patterns or actively interferes with surveillance equipment.
— —
