New 'FlyTrap' Method Disables AI Drones

The "FlyTrap" method exploits vulnerabilities in camera-based, autonomous target-tracking tech used in AI drones. By using specifically designed visual patterns on an ordinary umbrella, the system deceives the drone's neural network. The umbrella's patterns make the drone think the holder is moving farther away, causing it to fly closer. This continues until the drone is within capture range or crashes. This physical attack works without needing any wireless connectivity or external signals, and functions in various weather conditions. Researchers at the University of California, Irvine, developed FlyTrap, highlighting a critical security flaw with potentially far-reaching implications. Lead author Shaoyuan Xie and professor Alfred Chen suggest this is especially important as AI moves further into physical systems. They tested it on commercial drones like the DJI Mini 4 Pro. Unlike other counter-drone systems, FlyTrap can completely neutralize drones through physical capture or collision, not just redirect them. While effective for defense, there are concerns that it could be used to evade law enforcement or neutralize border patrol drones.