AI Safety Guardrails Built For Chatbots Fail To Protect Humans From Robots
Researchers from the University of Pennsylvania, Carnegie Mellon University and the University of Oxford argue in a new Science Robotics paper that the safety frameworks developed for AI chatbots are fundamentally inadequate for robots operating in the physical world, and that a layered, context-aware approach to robot safety is urgently needed. The paper, published April 29, focuses on th