Researchers Hack Autonomous Robots, Exposing Critical Vulnerabilities | JQJO

University of Pennsylvania researchers successfully hacked autonomous systems, including a self-driving car, a wheeled robot, and a quadrupedal robot. Using deceptive commands, they manipulated the car into driving off a bridge, the wheeled robot into identifying a bomb detonation site, and the dog robot into entering a restricted area. This highlights vulnerabilities in AI safety protocols.