Tag
This paper investigates how to engineer autonomous intelligent agents that can responsibly refuse user requests, anchoring non-compliance in justifications, override pathways, and tracking security risks and liability transfers.