This annex documents recurring governance design mistakes and operational failure modes observed across software and AI-enabled systems. Its purpose is to help architects recognize and avoid patterns that undermine controlled execution.
Understanding why governance fails is as important as understanding how it succeeds.
| Anti-Pattern | Description | Potential Mitigation |
|---|---|---|
| Authority Bypass | Execution proceeds without validating authority. | Mandatory authority gateway before consequence. |
| Implicit Trust | Systems assume trusted actors without verification. | Explicit authentication and authorization. |
| Missing Evidence | Decisions rely on incomplete supporting information. | Evidence admissibility checks. |
| Mutable Audit Records | Historical records can be altered. | Append-only ledger with integrity verification. |
| Policy Drift | Policy changes occur without governance oversight. | Versioned policies and controlled deployment. |
| Replay Gaps | Incomplete event history prevents reconstruction. | Replay completeness validation. |
| Unlimited Overrides | Emergency mechanisms become routine. | Time-bound, justified, and audited overrides. |
| Tight Coupling | Governance logic embedded directly in business code. | Policy-as-code and runtime separation. |
| Silent Configuration Changes | Operational settings change without traceability. | Configuration versioning and audit logs. |
| Unbounded AI Autonomy | AI performs consequential actions without governance. | Authority validation and human review where appropriate. |
Following a governance failure, organizations should preserve evidence, verify ledger integrity, replay affected decisions where possible, identify the root cause, and implement corrective controls before resuming normal operations.
Recognizing anti-patterns early reduces operational risk and improves the resilience, explainability, and auditability of governed execution systems.
Next: Companion Annex R — Migration Guide.
This document forms part of the CEI publication record. Supporting evidence is organized through the engineering-evidence archive, specifications, architecture documentation and replay verification materials.