Can Your Clothes Really Break AI Surveillance? How Hackers Are Outsmarting Facial Recognition
LAS VEGAS—Once upon a time, we gave away our faces and fingerprints for quicker, more secure access to our devices, and now, in 2026, we live in a surveillance-heavy society where license plate cameras on every corner can identify you using facial recognition technology.
These cameras not only know who you are but can also track your emotions, wardrobe, and daily habits. The companies behind these cameras collect and store massive amounts of data, prompting some law enforcement officers to use those cameras to look for just about anything or anyone for virtually any reason.
We can't claw back our data now, so we should try the next best plan: Break facial recognition methods by any means necessary. At the Black Hat security conference, one researcher demonstrated a way to fool current facial recognition models with a little help from the fashion world. It turns out, a scarf or a shirt could become your ticket to reclaiming some anonymity in public.
Beating Facial Recognition Tech: No Electronics or Curious Stares Required
Currently, defeating facial recognition technology requires an elaborate disguise in the form of a Batman-esque mask, elaborate makeup, infrared LEDs, or active electronics, all of which could invite unwanted attention.
Bill Swearington, a self-described hacker and founder of the SecKC cybersecurity community in Kansas City, developed an algorithm that creates adversarial patterns that, in the future, could be printed on fabric to cause facial recognition models to fail if you're wearing it.
As Swearington noted during his presentation, anti-facial recognition systems are not a new concept. In 2010, privacy artist and researcher Adam Harvey developed a system to counter burgeoning facial recognition technology, using adversarial makeup and hair-styling techniques.
After Swearington noticed hundreds of facial recognition cameras in his hometown of Kansas City, he decided to try to defeat the tech using a mix of modern and traditional methods.
How noRecognition Patterns Work
The technology used to create these patterns is called noRecognition, and Swearington has tested it against eleven different facial recognition models, including those running inside Palantir systems, Axon body cameras, and Clearview AI.
Swearington said that he's been working on generating and breeding adversarial patterns for about a year. While the tech isn't ready for prime time yet (it's only been proven to work digitally, not on physical fabrics), in the future, you can theoretically wear a scarf or a shirt printed with one of these adversarial patterns, avoiding AI detection altogether.
The research and testing are still in the early stages, and Swearington didn't reveal all the patterns he generated because, understandably, he doesn't want the camera companies to train their models on his data. If you're interested in his research, you can back the project on Kickstarter or check out his testing on his website .
PCMag and Yahoo may earn commission from links in this article.
