{
  "title": "The Authority Audit",
  "slug": "authority-audit",
  "objectNumber": 3,
  "version": "1.0",
  "status": "published",
  "source": "Questions: Series Three · Article 1. Failing-answer diagnostics: internal research, EXA Agentic Workflow Step 2, §1.2, ratified 2026-08-11.",
  "summary": "A self-administered, five-question diagnostic that maps where binding AI-behavior decisions actually get made in an organization, and who holds the standing to make them — paired with concrete 'what a failing answer sounds like' phrasing for each question, the instrument's actual diagnostic content. A structured companion to The EXA Lexicon's Authority Audit term (Standard object 1), which carries the five questions in prose-summary form only.",
  "lastUpdated": "2026-08-17",
  "changelog": [
    {
      "date": "2026-08-17",
      "version": "1.0",
      "note": "Published. Status/version flip only — no content changes in this entry, matching Career Ladder's and the EEA Rubric's own draft-to-published precedent. The object's content (the five questions, their failing-answer phrasing, the dual-source citation, and both judgment calls documented in this file's own header comment) is unchanged from its last authoring pass; this is the object's first published release."
    },
    {
      "date": "2026-08-16",
      "version": "0.1",
      "note": "Initial authoring, staged as draft per every prior object's own practice of author review before a page goes live. Built as a standalone object (Judgment Call 1) rather than an embedded Lexicon block, and rendered as a static question/failing-answer list with no scoring (Judgment Call 2) rather than the EEA Rubric scorer's pass/fail pattern — both decisions reasoned and flagged for author review in this file's own header comment. The five questions and their failing-answer phrasing are adapted from EXA_AgenticWorkflow_Step2_ImplementationPlaybook_v1.0.md, §1.2, an internal research document, verified against both that source and the Lexicon's existing prose summary before writing. Cross-linked to the Lexicon's existing Authority Audit term in both directions — this object's intro paragraph links back to the Lexicon entry, and the Lexicon's own definition gained one added sentence pointing forward to this object (additive-only, no Lexicon version bump, same commit). Not yet published: pending author review of this object's own editorial framing and both flagged judgment calls."
    }
  ],
  "type": "diagnostic",
  "tagline": "Five questions. The failing answer to each one is the diagnosis.",
  "authorityAudit": {
    "scopeNote": "Scope it correctly: run it where blast radius is widest — the autonomous, high-consequence surfaces, not every internal copilot.",
    "questions": [
      {
        "question": "Who ratifies a new agent's constraint set before it ships? Not who reviews it — who has standing to say this doesn't ship and have it hold.",
        "failingAnswer": "\"It goes through design review.\""
      },
      {
        "question": "Who can halt a deployment that violates the constitution?",
        "failingAnswer": "\"Escalate to the CTO and hope\" — that is a favor available on a good day, not standing halt authority."
      },
      {
        "question": "Who is in the room when compliance and legal set AI-behavior policy?",
        "failingAnswer": "Any answer that does not name design. Downstream reviews are not the room."
      },
      {
        "question": "Who owns the constitution when its author leaves?",
        "failingAnswer": "\"Priya wrote it\" — a specification dependent on one person's memory is tribal knowledge with a departure date."
      },
      {
        "question": "Whose performance review is affected when an ungoverned surface causes an incident?",
        "failingAnswer": "The sharpest question. It reveals whether accountability and authority sit in the same seat, or whether the org has split them so the person who takes the blame never had the power to prevent it."
      }
    ],
    "readingTheResult": "If most answers are \"nobody, formally\" or \"it depends who's around,\" the organization carries The Governance Vacancy — not a talent gap, not a process gap a better Jira workflow closes, but a seat that was never built."
  }
}