Advertisement
Tech by VICE

This Hacker Made Clothes That Can Confuse Automatic License Plate Readers​

Designer Kate Rose presented her "adversarial fashion" line of clothing, which introduces garbage data into license plate reader systems, at DEF CON 27.

by Samantha Cole
Aug 15 2019, 2:08pm

Images via Adversarial Fashion / Kate Rose

In a future where everything is surveilled by computer vision algorithms, even the patterns on our clothing can be subversive.

Digital security professional and fashion designer Kate Rose doesn’t just design textile patterns. She sullies the databases gathered by license plate readers by creating clothes covered in fake license plates.

If you've ever gotten a traffic ticket in the mail—sometimes complete with a photo of your vehicle, with yourself and passengers looking stunned in the front seat—your car has likely encountered an Automatic License Plate Reader, or ALPR.

Like many surveillance technologies in American cities, ALPRs are everywhere, always watching. They're attached to everything from public telephone poles to cop cars, but they're much more than speed trap cameras. ALPRs are privately-owned systems that capture everything resembling a license plate within their purview, collecting up to a thousand plates per minute. So while you're driving around, or traveling in a rideshare, these systems read the vehicle's plates and collect its GPS location and registration information, as well as the date and time.

According to the Electronic Frontier Foundation, ALPRs aren't just the crime-solving technologies law enforcement and vendors bill them as: They're also detailed surveillance systems, allowing police to track people as they travel to and from sensitive places like rehab centers and immigration clinics.

An example of adversarial fashion mimicking license plates
Credit: Kate Rose

But as advanced as these technologies are, they're easily confused by patterns and images that look just close enough to the real thing. "When an ALPR looks out into the world, it looks for something in the range of the camera that is likely to be a rectangle or a combo of letters that says it's a license plate," Rose said. The system crops that part of the image, and runs it through a database to compare to a training set of license plate examples. If the system determines it to be a plate, it uses Optical Character Recognition (OCR) to read, recognize, and save that plate, along with its GPS location and the time it was observed.

Rose's designs mess with that process. She developed and tested the repeating patterns on her fabrics to look as close to a real plate as possible, so that the system will save it as real. "By wearing these designs on the street, you help introduce junk data into a surveillance system that could in a large scale make it less useful, more expensive to use," she said. "I think it's important to experiment with ways to be noncompliant with surveillance systems in whatever way we can, and I like clothing as an accessible first step."

Rose presented her work during a talk at this year's DEF CON, a hacking conference held annually in Las Vegas. This year, DEF CON saw a 10 percent increase in the number of women in attendance, as well as several all-women and non-binary panels, presentations, and meetups throughout the conference—despite an ongoing history of misogyny in some DEF CON adjacent spaces.

While Rose's designs are gender-inclusive, as a woman in this industry, she's noticed more inclusivity in the conference as a whole, and a desire to fight back against oppressive systems of power.

"Hacking is fundamentally about democratizing and demystifying the systems and devices we all use, helping each other learn that we're smart enough to take apart and tinker with them, and reapply them to what you personally think is a good idea," she said.

This article originally appeared on VICE US.

Tagged:
defcon
algorithms
Automatic License Plate Reader
ALPR
def con 27