AI-defeating patterns block surveillance cameras from detecting people
Original: This ‘adversarial’ pattern can prevent surveillance cameras from detecting you
Why This Matters
Adversarial pattern research highlights the growing tension between AI-powered surveillance expansion and individual privacy rights.
Security researcher Bill Swearingen developed a computer-generated pattern called noRecognition that prevents surveillance cameras and license plate readers from detecting people and vehicles. After 31 million tests, the system was publicly demonstrated at Def Con 2026 in Las Vegas on August 9.
Bill Swearingen, a cybersecurity professional and co-founder of Kansas City meetup SecKC, spent over a year running approximately 31 million automated tests to develop adversarial patterns capable of defeating AI-powered surveillance detection. His project, noRecognition, generates on-demand patterns that—when printed on clothing or objects—prevent commonly deployed license plate readers and surveillance cameras from triggering detection alerts.
The patterns do not block cameras from recording footage. Instead, they disrupt the underlying object-detection algorithms, making the covered person or vehicle invisible to automated flagging systems while still appearing on video. Swearingen described the goal as allowing people to 'opt-out of being tracked,' calling privacy 'a fundamental right.'
The project received its first public demonstration on August 9 at the Def Con cybersecurity conference in Las Vegas, where Swearingen successfully printed the pattern on a vehicle and showed it could defeat real-world surveillance detection. Swearingen stated his motivation partly stemmed from wanting to attend a protest but feeling uncomfortable given the density of cameras capable of tracking individuals exercising constitutional rights to free expression.