GRCLibrary.ai

Principle

Fail-Closed Automation Principle

This document states, in citable adopter-facing form, a foundational design principle for automated systems: automation fails closed. When an automated system encounters an error, an ambiguous input, or the loss of a condition its correct operation depends on, it enters a known, controlled, safe state rather than continuing into an unsafe or permissive one. The corpus already enforces this piecemeal (network default-deny, privacy-by-default, fail-closed handling of AI errors), but never states it as a principle a control can cite. This document makes it citable and generalizes it across every automated boundary.

Type
Principle
Version
0.0.5
Date
2026-11-03
Owner
Governance Library Maintainer
Approving authority
Governance Library Maintainer
Classification
Public
Confidentiality
Public
Category
Core Governance
Review frequency
Annual and upon material governance, AI, or regulatory change
Licence
CC BY-SA 4.0

Read the full text on GitHub (opens in a new tab)

Outline

  • Purpose
  • Scope
  • 1. The principle
  • 2. The facets of fail-closed automation
  • 3. Fail-open as a deliberate, authorized exception
  • 4. Instantiation in the corpus
  • 5. Adoption guidance
  • Framework alignment

Frameworks it maps to

Related documents