Observability & Evaluation · Data artifact

Safety Metric Threshold Policy

Data artifactObservability & EvaluationObservability & Evaluationarc:SafetyMetricThresholdPolicy

A configuration of target values and action triggers for safety KPIs: violation rate, false-positive rate, human override rate and adversarial success rate.

Responsibility. Defines safety KPI targets and the actions their breaches trigger.

Also known as: Key safety indicators, Safety KPIs

Direct neighbourhood (hover for relationship types)

Relationships

configures structural

Guardrail Violation Monitor Ref9.01

Quantitative guidance

As stated by the sources; verify before use.

Safety violation rate target <0.1%; >0.5% triggers immediate review (Ref9.01).
False positive rate target <2%; human override rate target <10%; adversarial success rate target 0% (Ref9.01).

Classification

Quality attributes: Safety (ISO/IEC 25010 | NIST AI RMF: safe)

Sources

Ref9.01: "AI Safety Frameworks for Agent Systems," unpublished reference note (01-AI-Safety-Frameworks.md), Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam supplementary materials, 2026. unpublished note