Adversarial Fashion Designed to Trick Automated License Plate Readers

As it turns out, surveillance cameras that have been “trained” to spot and read license plates aren’t all that good at discerning real ones from fakes. That makes it pretty easy to trick Automatic License Plate Reader (ALPR) systems with images of fake plates, making it possible to flood their databases with unusable information.

When hacker and fashion designer Kate Rose learned – through a conversation with Dave Maass, a researcher with the Electronic Frontier Foundation – that the plate readers kind of suck at their jobs, she got an idea. Her new line “Adversarial Fashion” is the result. Unveiled at the DefCon cybersecurity conference in Las Vegas last week, the garments spell out the words of the fourth amendment of the US constitution, which protects Americans from “unreasonable searches and seizures.”

The cameras, however, read the garments as real license plates, and the proof is in their databases. ALPRs are always on, and can collect thousands of plates per minute, so for the system, there’s nothing unusual about capturing so many individual plates at a time. As Rose’s presentation at DefCon noted, overloading this kind of surveillance technology is one of the main methods of confounding it (along with blocking the collection of information.)

If you’re interested in making your own, Rose has provided all of the information you need to do it. But her line of Adversarial Fashion is pretty affordable, with prices starting at $24.99 – check it out here, or follow the brand new Instagram account @adversarialfashion.