Admissibility and Standing in Artificial Epistemic Agents

This paper establishes standing as a necessary condition for admissible reasoning in artificial epistemic agents and proves its strict equivalence to admissibility under fail-closed evaluation. It shows that agents lacking standing cannot produce meaningful, reference-grounded claims, regardless of internal coherence, learning capacity, or optimization success. The analysis is necessity-based and exhaustion-driven: all proposed alternatives to standing (probabilistic validation, performance metrics, post-hoc justification, governance overlays, or repair strategies) are shown to fail structurally. No algorithms or construction procedures are provided. The result is a binary boundary on what counts as a legitimate epistemic agent and a foundational constraint on artificial reasoning systems.