PCS-01: Runtime Governance Survivability Diagnostics for Operational AI Systems — A Simulation-Based Architecture for Drift, Escalation, and Authority Failure Analysis

Fuente: Zenodo
Enregistré dans:
Détails bibliographiques
Auteur principal: Senke, Alexanja
Format: Recurso digital
Publié: Zenodo 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866901945748291584
author Senke, Alexanja
author_facet Senke, Alexanja
contents <div> <p class="MsoNormal"><span>Current AI governance frameworks evaluate compliance, auditability, and model behavior in isolation from operational conditions. Operational AI systems, however, increasingly fail through structural degradation that remains invisible to traditional governance instruments — they drift, delegate incorrectly, and continue executing after governance capability has effectively collapsed.<span>  </span>This paper introduces PCS-01 (Provider Compliance Simulator), a runtime governance stress environment designed to simulate and observe governance survivability under operational conditions. Rather than evaluating whether AI outputs are correct, PCS-01 evaluates whether governance remains executable during runtime.<span>  </span>The architecture models five primary failure modes: Drift Accumulation (FM-01), Escalation Failure (FM-02), Carrier Loss (FM-03), Recursive Delegation (FM-04), and Silent Corruption (FM-05). PCS-01 introduces the concept of Governance Survivability as a measurable runtime property, distinct from auditability and policy compliance, and operationalises it through a four-gate Admissibility Engine (Q1–Q4), a Human Commit Boundary (HCB), and a three-level recovery architecture (R1–R3).<span>  </span>The system is positioned as a runtime diagnostic architecture for operational AI systems in telecommunications, healthcare, critical infrastructure, and other high-consequence regulated environments operating under the EU AI Act (2024) and comparable governance frameworks. A companion software demonstrator (PCS-01 Runtime MVP) is published separately on Zenodo.</span></p> </div>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_20331819
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle PCS-01: Runtime Governance Survivability Diagnostics for Operational AI Systems — A Simulation-Based Architecture for Drift, Escalation, and Authority Failure Analysis
Senke, Alexanja
AI Governance · Runtime Governance · Governance Survivability · Operational AI Systems · Drift Detection · Escalation Integrity · Human Oversight · EU AI Act · Admissibility · Authority Continuity · Human Commit Boundary · Provider Compliance · Telecommunications AI · AI Safety · Simulation Architecture
<div> <p class="MsoNormal"><span>Current AI governance frameworks evaluate compliance, auditability, and model behavior in isolation from operational conditions. Operational AI systems, however, increasingly fail through structural degradation that remains invisible to traditional governance instruments — they drift, delegate incorrectly, and continue executing after governance capability has effectively collapsed.<span>  </span>This paper introduces PCS-01 (Provider Compliance Simulator), a runtime governance stress environment designed to simulate and observe governance survivability under operational conditions. Rather than evaluating whether AI outputs are correct, PCS-01 evaluates whether governance remains executable during runtime.<span>  </span>The architecture models five primary failure modes: Drift Accumulation (FM-01), Escalation Failure (FM-02), Carrier Loss (FM-03), Recursive Delegation (FM-04), and Silent Corruption (FM-05). PCS-01 introduces the concept of Governance Survivability as a measurable runtime property, distinct from auditability and policy compliance, and operationalises it through a four-gate Admissibility Engine (Q1–Q4), a Human Commit Boundary (HCB), and a three-level recovery architecture (R1–R3).<span>  </span>The system is positioned as a runtime diagnostic architecture for operational AI systems in telecommunications, healthcare, critical infrastructure, and other high-consequence regulated environments operating under the EU AI Act (2024) and comparable governance frameworks. A companion software demonstrator (PCS-01 Runtime MVP) is published separately on Zenodo.</span></p> </div>
title PCS-01: Runtime Governance Survivability Diagnostics for Operational AI Systems — A Simulation-Based Architecture for Drift, Escalation, and Authority Failure Analysis
topic AI Governance · Runtime Governance · Governance Survivability · Operational AI Systems · Drift Detection · Escalation Integrity · Human Oversight · EU AI Act · Admissibility · Authority Continuity · Human Commit Boundary · Provider Compliance · Telecommunications AI · AI Safety · Simulation Architecture
url https://doi.org/10.5281/zenodo.20331819