The world knows a lot more about your face than you realize. Your image appears in surveillance camera footage from city streets, gas stations, and Target. If you have a phone with Face ID, Apple knows the intimate details of what makes your mug unique. Facebook, too, knows quite a bit about your face from pictures you and your friends have uploaded (and it’s facing lawsuits for illegally collecting and storing that data). In some cases, your face is your boarding pass. Children’s summer camps even use cameras with facial recognition to send notifications and photos to campers’ parents. Soon, cameras might be able to do more than identify your face: Amazon says its Rekognition software can now analyze faces for emotions like fear, anger, and surprise.
The Chinese government is particularly adept at monitoring its citizens, and it is not shy about using facial recognition to track minority groups and issue traffic citations. It plans to give companies access to the technology, for targeted marketing and giving discounts to attractive people. As anti-government demonstrations have raged in Hong Kong over the last several months in response to a bill calling for HK residents to be extradited to mainland China, protesters have taken steps to ensure those cameras won’t record their faces for China’s growing facial recognition database. Stores reported huge sales of helmets, masks, and goggles as protesters steeled themselves against police attacks, but those accessories also served a secondary purpose: avoiding detection by cameras. Protesters have also flashed laser pointers to daze cameras and police on the scene.
As facial recognition tech takes hold in more parts of our daily lives, eluding surveillance may be desirable even if you aren’t engaged in protesting. You might want to be able to hop over to the grocery store without an N95 mask obscuring half your face to thwart recognition technology. Enter anti-surveillance fashion.
Some designs, like these bold hairstyles and geometric makeup patterns that evade algorithms, look most at home on a catwalk model. If you were to wear that look out in public, you might be fool cameras, but you’d also attract attention from passersby, as the Atlantic’s Robinson Meyer learned when he used such methods for a few days in 2014 while going about his life in Washington, D.C. “The very thing that makes you invisible to computers makes you glaringly obvious to other humans,” he wrote.
Just like high-fashion looks at runway shows aren’t meant to be worn by the average person going to their average-person job, many of these anti-surveillance looks serve more as art pieces. “The main significance is creating awareness,” says Henry Navarro Delgado, an associate professor of fashion at Ryerson University in Ontario. “That’s why fashion is so effective: You have something to say, you wear it, people see you, it’s immediate. Part of the purpose is to make people who normally don’t think about this aware that these technologies are out there, and we’re being watched.” Navarro also told me that “[t]here’s a long-standing history of people using our appearance to spread our beliefs.”
One of the earliest examples of this goes back to the 5th century B.C., when Athens and Sparta were dueling in the Peloponnesian War. Within Athens, a Spartan-sympathetic faction adopted the more conservative dress styles popular in Sparta to signal their political leanings. More recently, activists like the Black Panthers adopted a signature style—black leather, combat boots, beret—that’s been adopted and modernized by antifa protesters. And certain items of clothing, like pantsuits and pussy hats (remember those?), have become symbols in some feminist circles.
Anti-surveillance styles aren’t what Navarro classifies as “fashion.” We may use the word colloquially to refer to anything connected with clothing and appearance, but Navarro emphasizes the distinction between a specific type of garment or styles worn by a few people and the cultural institution of fashion. “Fashion is when society at large reaches an agreement that something is in style,” says Navarro. That means it’s tied to a specific era and cultural moment, from which it derives meaning. “Fashion is like food: It comes with an expiration date.”
Artists, designers, and computer vision experts are creating the principles behind anti-surveillance couture. Currently, there are two main strategies to confuse surveillance algorithms. The first is to block it from collecting any data. That’s what the Hong Kong protesters’ masks and lasers accomplish: By obscuring facial features or incapacitating the camera’s ability to record, the software can’t run its usual analyses. The other method is to overload algorithms with junk data. If the current anti-surveillance garments become more popular, you might see the same design elements incorporated into everyday clothing.
That’s what typically happens with runway looks or conceptual designs: They serve as inspiration for more mainstream looks. (Take low-rise jeans, peplum, or any other trend from the past 20 years or so. Most got popularized in high-fashion circles, then made their way to TJ Maxx.) Similarly, political dress is often co-opted into fashion, but in a watered-down, more palatable way. “The Black Panthers always wore leather, and it was always black, but when that got to the mainstream, people started wearing browns and earth tones, which softened the message a little,” says Navarro. “Mainstream fashion takes cues from fringe movements and repurposes it in a way that’s a little friendlier.”
Already, there are more subtle anti-surveillance styles. Adversarial Fashion’s T-shirts, bomber jackets, hoodies, and crop tops come in two prints designed to inject junk data into license plate readers’ databases. As you might guess, both designs incorporate what looks like a bunch of license plates. One pattern just looks like a scattering of custom California plates with gibberish vanity tags (“NTS SHAL,” “S OR THI,” “ECTS AG”) but, if you read it like a human would, it’s actually the text of the Fourth Amendment, which prohibits unreasonable searches and seizures. The other also incorporates faux plates but with an artsier-looking circuit board pattern. Adversarial Fashion advises that for “maximum effect,” customers should consider sizing up, because shirts need to hang straight down for the fake plate numbers to be read by cameras.
Another pattern, called Hyperface, also relies on overloading surveillance cameras with junk data, but with fake faces instead of license plates. From far away, it looks like a collection of little squares, but to a facial recognition algorithm, the abstract squares look like dozens of little faces. It can be printed on everything from leggings to shower curtains, but your cheapest option is a Hyperface sticker, which costs only $2.47.
CV Dazzle’s hair and makeup styles is one of several projects meant to be part of an open-source movement sharing what’s effective at throwing off algorithms. Even those trying to make money see themselves as part of a political movement. Adversarial Fashion, which sells clothing starting at $24.99, provides a list of resources and tutorials for people (Hong Kong protesters, perhaps?) to create their own designs that confuse computer vision, which can then be printed on whatever you want. Crucially, the company makes no claims about the overall efficacy of its design. In a presentation, the company says that because “government surveillance contractors and agents are no[t] going to expose access to their systems,” it has no access to training data or algorithms to test its products against the real thing. In all likelihood, researchers are already working on computer vision methods to outsmart popular anti-surveillance designs. If anything, private individual designs may stand a better chance against algorithms, just because they’re not publicly available as test cases for training anti-anti-surveillance techniques.
Wearing these more understated styles, you wouldn’t look out of place as you rode public transit or went out to dinner, but you’d still most likely evade detection—and you send the signal that you’re opposed to the lack of privacy. Other trends Navarro can see becoming more mainstream are shiny, reflective finishes, which can look glamorous (think silver or gold sequins) but may also bounce light around in a way that disrupts cameras’ ability to read a face.
Elements of camo, too, might be used to obscure outlines. It seems only fitting that camo, a strategy refined and heavily used for military use, now serves as inspiration for countersurveillance measures for citizens to use against the military. As surveillance methods catch up to counteract our initial anti-surveillance styles, those styles will just have to evolve again. “It’s an arms race,” says Navarro,” and these [anti-surveillance] principles will become integrated into mainstream styles, because surveillance is not going to go away.”