Generative AI Agents in Autonomous Machines: A Safety Perspective