analiza aiqbench
This paper addresses the "assurance-transition gap" by proposing Runtime Assurance Contracts (RAC), a formal schema that binds agent autonomy to real-time evidence states rather than relying solely on aggregate performance scores. Through deterministic failure-injection studies in agentic coding, the authors demonstrate how mandatory gates can prevent high-risk actions that score-based rules would erroneously permit, highlighting critical safety mechanisms for deploying autonomous agents in clinical and judicial domains.
Trust Is Not a Score: Runtime Assurance Contracts for High-Risk AI Agents
arXiv:2609.39717v1 Announce Type: new Abstract: Benchmarks, audits, and agent protocols describe performance, permissions, and repair, but not how observed evidence should change an agent's authority during a consequential task. We call this the assurance-transition gap. We propose a Runtime Assurance Contract (RAC), a policy-level formal schema binding autonomy boundaries, component eligibility, evidence state, transition policy, human-review capacity, and non-compensatory gates. Under RAC, soft metrics may inform routing, whereas a failed or unknown mandatory gate forces retry, switch, escalation, deferral, or stop; aggregate performance cannot authorize action. We define the contract, an evidence record, a permission rule, and five invariants, and illustrate them in clinical, industrial, and judicial failure probes. We then report a deterministic failure-injection study in agentic coding: 280 constructed cases evaluated by a gate conjunction, a score-only rule, and a restricted protocol baseline. At the published example weights and threshold, the score rule admits 80 of 100 block-required injections and all 40 review-required injections. Tuned in hindsight, it matches the conjunction on this corpus. For positive weights, a positive threshold, binary risk signals, zero-signal controls, and an injected case firing each signal alone, we show that exact agreement holds if and only if the threshold does not exceed the smallest weight. A separate set of 18 hand-authored traces checks version-pinned evidence and review transitions against simpler policy variants. In a further prospective synthetic holdout of 24 episodes, two blinded LLM judges assign identical labels to all 72 action attempts; RAC and a separately implemented full stateful baseline both match these labels. These studies test mechanisms on synthetic cases; they establish neither deployed safety nor cross-domain effectiveness.
Abstract:Benchmarks, audits, and agent protocols describe performance, permissions, and repair, but not how observed evidence should change an agent's authority during a consequential task. We call this the assurance-transition gap. We propose a Runtime Assurance Contract (RAC), a policy-level formal schema binding autonomy boundaries, component eligibility, evidence state, transition policy, human-review capacity, and non-compensatory gates. Under RAC, soft metrics may inform routing, whereas a failed or unknown mandatory gate forces retry, switch, escalation, deferral, or stop; aggregate performance cannot authorize action. We define the contract, an evidence record, a permission rule, and five invariants, and illustrate them in clinical, industrial, and judicial failure probes. We then report a deterministic failure-injection study in agentic coding: 280 constructed cases evaluated by a gate conjunction, a score-only rule, and a restricted protocol baseline. At the published example weights and threshold, the score rule admits 80 of 100 block-required injections and all 40 review-required injections. Tuned in hindsight, it matches the conjunction on this corpus. For positive weights, a positive threshold, binary risk signals, zero-signal controls, and an injected case firing each signal alone, we show that exact agreement holds if and only if the threshold does not exceed the smallest weight. A separate set of 18 hand-authored traces checks version-pinned evidence and review transitions against simpler policy variants. In a further prospective synthetic holdout of 24 episodes, two blinded LLM judges assign identical labels to all 72 action attempts; RAC and a separately implemented full stateful baseline both match these labels. These studies test mechanisms on synthetic cases; they establish neither deployed safety nor cross-domain effectiveness.
16 pages, 2 figures, 5 tables. Ancillary files: decision log, executable transition model, LLM-labelled synthetic holdout. Synthetic mechanism study; no deployment claim
Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Software Engineering (cs.SE)
arXiv:2609.39717 [cs.AI]
(or arXiv:2609.39717v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.39717 arXiv-issued DOI via DataCite (pending registration)
From: Serhii Zabolotnii Dr. [view email] [v1] Wed, 30 Sep 2026 13:28:42 UTC (232 KB)
Text integral sub licența deschisă a publicației · 325 cuvinte
Subiecte conexe
Presentation: Exploding GPUs
InfoQ · 16 oct. 2026
Vertical Advantage: Transforming Industries with Lakebase and Agentic AI
Databricks · 27 aug. 2026
A Red Line and Oversight Framework for Government AI Contracts
AI Alignment Forum · 18 iul. 2026
Digital States 2026: AI Moves From Experiment to Infrastructure
govtech.com · 24 sept. 2026