SOURCE-LINKED INTELLIGENCE

Approved Too Late: Verdict Staleness in LLM-Guarded Self-Adaptive Systems

arXiv · AI, language, vision and robotics · article · Aug 26, 2026 · UTC

A large language model (LLM) guardrail for a self-adaptive system (SAS) may issue an approval that is correct at check time but stale by actuation. This creates an Execute-stage time-of-check to time-of-use (TOCTOU) hazard. We study verdict freshness: whether a guardrail verdict remains valid when used. We distinguish three quantities that answer different questions: all-candidate verdict change under fixed-action replay, oracle-labeled approval expiry on recorded closed-loop trajectories, and judge-conditioned use-time invalidity. Across five reproducible SAS environments, all-candidate verdi

Read original source ↗ Open in workspace

recordType

paper

region

Global