Governance & Compliance · Data artifact

Hallucination Threshold Policy

Data artifactGovernance & ComplianceSafety, Security & Governancearc:HallucinationThresholdPolicy

A policy artifact setting maximum acceptable hallucination rates per severity level and application domain for deployment and alerting decisions.

Responsibility. Declares domain- and severity-specific hallucination tolerances.

Also known as: Domain-specific hallucination thresholds, Severity thresholds

Direct neighbourhood (hover for relationship types)

Relationships

configures structural

constrains control

Rollout Manager abstract Ch3.10

Design guidance

MUST set thresholds by failure consequence per domain and severity rather than one aggregate target.

Quantitative guidance

As stated by the sources; verify before use.

Severity thresholds: critical <0.1-0.5%, high <1-2%, medium <5%, low <10-15% (Ch3.10).
Domain thresholds: financial advice <2%, customer support <5%, healthcare critical <0.5%, entertainment <20% (Ch3.10).
Production baseline hallucination rates range 8-20%; customer-facing targets <5% (Ch3.10).
Alert when hallucination rate exceeds 2% (Ref8.03).

Classification

Quality attributes: Safety (ISO/IEC 25010 | NIST AI RMF: safe)Interaction capability (ISO/IEC 25010)
Risks mitigated: Aggregate rates masking critical hallucinations

Sources

Ch3.10: T. Nguyen, "Efficiency Metrics," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.10. ISBN: 9798244538229.
Ref8.03: "Agent Evaluation Frameworks and Metrics," unpublished reference note (03-Agent-Evaluation-Frameworks.md), Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam supplementary materials, 2026. unpublished note