Hacker Wraps Toyota in Weird Pattern to Blind Flock AI Police Cameras

At the DEF CON security conference in Las Vegas, researcher Bill Swearingen rolled up with a 2009 Toyota Yaris completely wrapped in an AI-scrambling pattern. The adversarial vinyl wrap successfully turned the car invisible to roadside automated surveillance, making smart readers completely miscalculate or ignore its presence.

To craft the optical illusion, Swearingen trained a reinforcement learning algorithm specifically to locate blind spots in object recognition neural nets. The resulting visual noise baffled 11 open-source machine vision models, roadside license plate readers made by Flock, and even Axon police body cameras.

The project grew into a wider anti-surveillance effort called noRecognition, offering hoodies and shirts printed with the same machine-breaking geometry, with vehicle covers coming next. Meanwhile, pushback against municipal camera grids has turned chaotic, with residents physically tearing down roadside poles, a military technician facing charges for felling dozens of units, and journalists getting mistakenly swarmed by cops over AI false positives.

Under growing heat, Flock quietly tweaked its terms by cutting default footage retention down to a single week while adding a manual toggle for long-term storage.

When mass surveillance algorithms get paralyzed by a glorified car sticker, the entire multi-billion-dollar automated policing empire starts looking surprisingly fragile.

Source: TechCrunch