Adversarial clothing designed to confuse facial recognition could soon hit mainstream fashion

'Adversarial clothing': are garments designed to confuse facial recognition systems about to go mainstream?

Designers say that as well as offering a degree of protection from surveillance, their clothes make a powerful fashion statement about the importance of privacy As facial recognition technology is rolled out across Britain's public spaces, a new generation of designers say privacy could be the next big fashion trend. Companies have started incorporating "adversarial patterns" in their garments - carefully designed arrangements of shapes, colours and repeated motifs said to exploit weaknesses in some computer vision systems. The designers say advances in computing have made it easier to incorporate such patterns into commercially viable garments. Experts caution that the effectiveness of the patterns depends on the surveillance system and the conditions in which it is used, but Nick Tidball, the co-founder of the clothing brand Vollebak, thinks "adversarial clothing" could be on the cusp of going mainstream. "Anti-surveillance feelings are so widespread that all it would take is for a single celebrity to wear one of these garments, currently popular in the countercultural fashion world, to a high-profile event for it to take off," he said. "So-called 'adversarial clothing' wins on many levels. As well as the practicality of protection, it's fashionable and fun, it makes a powerful, public statement that many are in agreement with, it spreads even more awareness about the importance of privacy and it helps encourage public debate." Unlike traditional CCTV, modern computer vision systems can identify faces, follow individuals across cameras and search footage at scale. Recent advances in generative AI have made this type of automated identification cheaper and more widely available to police, retailers and private businesses, an expansion recently warned against by Britain's biometrics watchdogs, which have called for more laws and a regulator to clamp down on misuse. Evidence of misuse and that black and Asian people are more likely to be incorrectly identified than white people has led to increasing public concern. A recent poll showed almost 60% of people believed facial recognition was "another step towards turning the UK into a surveillance society". Dr Jennifer Bell, a senior lecturer specialising in creative AI, fashion and digital culture at Nottingham School of Art & Design, said clothing with anti-facial recognition designs was increasingly available at high street prices and was being marketed to a wide demographic. "That growing awareness combined with a lowering of cost often precedes the tipping point towards a real cultural moment," she said. Daniel Preuß, the co-founder of the Urban Privacy clothing brand, said new technology meant you could now "combine smart, striking style with invisible protection". He emphasised that because surveillance systems are so powerful, no design can guarantee security from detection, but said "the added value of fashion is to spread awareness and help propagate public discourse". Preuß said his designs used large-scale prints, asymmetrical cuts and streetwear-inspired silhouettes to confuse facial recognition algorithms. The company said its Urban Ghost coat integrates LEDs into the hood that emit infrared light to dazzle night-vision surveillance cameras. Preuß, who co-founded his company after reading about the whistleblower Edward Snowden's revelations about US surveillance in the Guardian, said his designs played with the fact that "facial recognition systems freak out when they see multiple faces at once". "Our patterns play with that chaos, confuse algorithms and make it way harder to pin you down," he said. Bell, however, said "none of these products are tried and tested, and a lot of these surveillance technologies can deal with a little resistance ... [but] even if the designs don't necessarily work perfectly, fashion is also a visible sign of resistance. "This is consumers collectively coming together to make a visible statement." Rachele Didero, the founder of Cap_able, which creates clothing marketed as making AI recognition more difficult, said interest in her brand had rocketed in the last few years. "When I started doing this in 2018, people thought I was designing masks to rob banks," she said. "But now these concerns are no longer niche. New generations are increasingly afraid of AI and concerned for their privacy," she said. "Those wearing these products are the vanguard. The mainstream is quickly coming up behind them, helped by the fact that bigger companies will see the potential for profit and will push the trend into the popular, public domain, changing the way we dress on a large scale." Tidball, however, said that whether anti-surveillance fashion became mainstream may ultimately depend less on designers than on governments. "If such clothing genuinely proved effective, it could get political very quickly," he said. "Then this type of clothing could find itself banned."

Anti-surveillance clothing is getting cheaper, but don't expect an invisibility cloak

Affordable shirts now claim to confuse facial recognition, although their protection depends heavily on the camera and software watching you Anti-surveillance clothing is starting to look less like an art-school experiment and more like something you could actually wear outside. Shirts designed to confuse facial recognition systems now cost about as much as ordinary streetwear, although buying one won't make you disappear. The Guardian reports that designers are using face-like prints, unusual cuts and infrared lights to interfere with computer vision. These techniques target specific weaknesses, so their success depends on what happens to be watching you. How can clothing confuse a camera Adversarial clothing takes advantage of the shortcuts software uses to identify people and objects. Urban Privacy's Faception designs scatter fake faces across the fabric, giving an algorithm more visual noise to process. Cap_able uses knitted patterns created with AI. The company says versions of the YOLO object-detection system have mistaken its designs for animals or small figures instead of identifying the wearer as a person. It's an amusing result, but fooling one model doesn't guarantee the same trick will work elsewhere. Recommended Videos Urban Privacy's experimental Urban Ghost coat tries a different approach. Infrared LEDs around its hood are intended to overwhelm compatible night-vision cameras. Regular cameras and other surveillance systems may remain completely unimpressed. How affordable is privacy fashion Urban Privacy lists Faception Reloaded T-shirts from €35, sweatshirts from €59 and hoodies from €65. At those prices, anti-surveillance clothing isn't reserved for wealthy privacy enthusiasts or gallery mannequins. Cap_able remains considerably more expensive. Its knitted crop tops start at €560, while a hoodie costs €620. That puts the collection firmly in wearable-art territory, but the cheaper Urban Privacy garments show how quickly the idea is filtering down. You can now test an adversarial pattern without making your bank account unrecognizable too. Why shouldn't you trust it completely Controlled testing against one object-detection model can't prove a garment will defeat facial recognition in public. Lighting and camera position can alter the result, while newer software may learn to ignore patterns that once caused trouble. Researcher Jennifer Bell also told The Guardian that these products haven't undergone independent real-world testing. That makes adversarial clothing more convincing as protest than protection. Wearing one makes a clear statement about surveillance and may inconvenience certain systems along the way. Just don't treat a patterned hoodie as an anonymity switch. Until these garments face broader independent testing, assume the camera can still see you.

These Clothes Were Designed to Trick AI Surveillance Cameras

As fears of surveillance rise thanks to security cameras in every public (or private) space, an unexpected form of resistance is popping up: adversarial fashion. This unique type of fashion is created with the intention of obscuring people from AI facial recognition software. These "adversarial" pieces of clothing and accessories employ special patterns and materials made to confuse AI facial recognition software. Sometimes they include unique patterns that can cause AI to misidentify people as animals or objects. Other anti-surveillance clothing technology uses infrared reflective gear that blocks faces out entirely on camera, as if a bright light was shining on them. There's a range of brands selling adversarial clothes at wildly different price points. On the more practical and accessible end, Anti-AI Clothing sells masks and gaiters under $20 and T-shirts for around $30. Looking to the other end of the spectrum, Italian-based company Cap_able sells a $780 reversible dress, one side facial recognition reads as a person and the other a zebra. While at the Fashion Institute of Technology in 2019, Cap_able's founder and CEO Rachele Didero began developing adversarial apparel after noticing just how many biometric cameras were in New York City.