AI RESEARCH
Detecting Is Not Resolving: The Monitoring Control Gap in Retrieval Augmented LLMs
arXiv CS.AI
•
ArXi:2605.27157v1 Announce Type: new Retrieval-augmented LLMs are deployed for tasks where evidence quality determines action safety, yet evaluation protocols assume that single-turn robustness predicts robustness when evidence accumulates across turns. We show this assumption is fundamentally incorrect. Models exhibit a monitoring-control gap: they readily acknowledge contradictory evidence, yet this awareness fails to constrain their final recommendations - detecting epistemic conflict does not imply resolving it safely.