AI RESEARCH

Why Agents Compromise Safety Under Pressure

arXiv CS.AI

ArXi:2603.14975v1 Announce Type: new Large Language Model agents deployed in complex environments frequently encounter a conflict between maximizing goal achievement and adhering to safety constraints. This paper identifies a new concept called Agentic Pressure, which characterizes the endogenous tension emerging when compliant execution becomes infeasible. We nstrate that under this pressure agents exhibit normative drift where they strategically sacrifice safety to preserve utility.