Cybersecurity researcher creates patterns designed to fool surveillance cameras
A US researcher spent a year running about 31 million tests to develop computer-generated patterns that can stop common surveillance cameras and license plate readers from detecting people or vehicles. The project had its first public demonstration at the Def Con conference in Las Vegas.
Kansas City-based cybersecurity professional Bill Swearingen has developed a project called noRecognition aimed at helping people avoid automated detection by modern surveillance cameras. He says he has run roughly 31 million tests over the past year to create patterns that, when printed on clothing or objects, prevent detection algorithms from recognizing whatever they cover.
The patterns don't stop cameras from recording footage, but they disrupt the algorithms typically used to identify faces, people, or vehicle license plates, preventing detection alerts from being triggered. Swearingen describes privacy as a fundamental right and frames the project as a way for people to "opt out" of being tracked.
From open-source algorithms to a self-training model
The project began with tests that gradually defeated individual open-source video detection algorithms. Over time, this evolved into a reinforcement learning model that trains itself by repeatedly generating patterns until they can beat multiple detection systems at once. The model can now defeat all 11 open-source detection algorithms tested, including software powering Flock license plate readers, Axon body-worn cameras, and Clearview AI.
Last Friday at the Def Con conference in Las Vegas, Swearingen ran his first real-world test with help from Donut Media, covering a 2009 Toyota Yaris in one of his newest patterns to see whether it could avoid detection by a Flock camera. He said the test was successful, though the car's wheels posed a challenge.
Swearingen said he is keeping his strongest patterns offline to prevent camera manufacturers from quickly countering them. The noRecognition project also runs a crowdfunding campaign to produce merchandise featuring the patterns, such as T-shirts and hoodies, with vehicle wraps potentially coming in the future.
