RFC-021: AIGP-SGL — Symbolic Governance Language for Agentic and Autonomous AI — 11. Default Evaluation Rules

AIGP Specification › RFC-021: AIGP-SGL — Symbolic Governance Language for Agentic and Autonomous AI › 11. Default Evaluation Rules

← 10. Decision Semantics · Section index · 12. Canonical AST →

11. Default Evaluation Rules

AIGP-SGL uses conservative defaults.

11.1 Default Deny

If no valid permit exists, the action is denied.

default(action) = deny

11.2 Default Inhibit for Force

For force-capable actions:

default(force_action) = deny
default(force_capability) = inhibited unless explicitly authorized

11.3 Safe State on Uncertainty

If governance state is uncertain, the system MUST enter the least harmful safe state available.

if governance_state == uncertain → degrade_or_stop

11.4 Deny on Invalid Authority

if authority.invalid ∨ authority.expired ∨ authority.unverified → deny

11.5 Deny on Broken Evidence Binding

if ddna_binding.invalid → deny_or_escalate

11.6 Stricter Rule Wins

When permit and deny conflict:

deny > permit
inhibit > permit
safety > mission_objective
human_protection > system_self_preservation

---

---

← 10. Decision Semantics · Section index · 12. Canonical AST →