AI RESEARCH

LoopTrap: Termination Poisoning Attacks on LLM Agents

arXiv CS.AI

ArXi:2605.05846v1 Announce Type: cross Modern LLM agents solve complex tasks by operating in iterative execution loops, where they repeatedly reason, act, and self-evaluate progress to determine when a task is complete. In this work, we show that while this self-directed loop facilitates autonomy, it also