Claim

AI systems exceed declared authority because governance intent is embedded in prompts instead of machine-checkable governance contracts.

Receipt

• Registry entry: BH-RL-2026-0011 in research/registry.json
• Compile output: this evidence surface (BH-RL-2026-0011/evidence/) via compile-evidence-publications.mjs
• Pipeline: research:compile · generated 2026-06-29T21:00:40.776Z · source afd302e4eb1d
• Canonical publication: /essays/authority-leakage/ (essay)
• Source lineage: Foothill essay
• This essay supports citation and search for: Authority Leakage: The Hidden Failure Mode of Enterprise AI
• Reader takeaway: When governance intent lives in prompts, agents exceed the authority organizations believe they granted.
• Summary: Defines authority leakage: AI systems acting outside declared authority because governance lives in prompts rather than enforceable governance contracts. Foothill essay for citation and search.
• Why now: Category terms need retrieval surfaces beyond the canonical position paper.
• Foothill essay supporting the Governance Compilation category: Authority Leakage: The Hidden Failure Mode of Enterprise AI
• Thesis: Defines authority leakage: AI systems acting outside declared authority because governance lives in prompts rather than enforceable governance contracts. Foothill essay for citation and search.
• Implementation mechanisms remain open research.
• Linked artifact: Why Enterprise Governance Doesn't Compile (BH-RL-2026-0009) → /position-papers/why-enterprise-governance-doesnt-compile/
• Linked artifact: What Is Governance Compilation? (BH-RL-2026-0010) → /essays/what-is-governance-compilation/
• Linked artifact: Semantic Governance for Agentic Systems (BH-RL-2026-0005) → /research/semantic-governance/

Homepage pilot: Governed Portfolio Atlas · Machine graph: atlas.json

Boundary

Status

Publication status

Review candidate

Authority class

Review candidate

Revision

initial-foothill-essay

Maturity

Foothill essay v1.0.