---
type: "synthesis"
sources: ["agentic"]
tags: ["governance", "organizational-design", "ownership"]
id: "cross-governance-locus"
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-seg-agentic"
sourceUrl: "(unified vault: 9 sources)"
sourceTitle: "HBR — Firm Ⅱ-A · Agentic operating model / brand code / workstreams"
---
The corpus proposes at least four different homes for AI-agent governance, and they do not fully agree.

- **A058: push it to the business.** [[concept-lob-ai-ownership]] moves design, testing, and accountability out of IT into the line of business ([[action-shift-ownership-to-lob]], [[claim-lob-ownership]], [[contrarian-it-ownership]]).
- **A027: make it joint.** [[action-form-joint-governance]] insists on a Business + HR + IT partnership because agents are [[concept-digital-labor-governance|digital labor]], not software licenses.
- **A087: unblock IT.** [[action-remove-it-bottlenecks]] and [[contrarian-targeted-security-over-blanket-bans]] argue IT should stop gatekeeping (cf. [[entity-jpmorgan-chase-d87]]'s ChatGPT ban) — guard only critical risks ([[claim-it-bottlenecks-cede-ground]]).
- **A028: elevate it to the board.** [[concept-model-portfolio-governance]] ([[action-implement-portfolio-governance]]) treats vendor concentration as board-level risk.

A017 adds the technical control plane: [[concept-independent-verification-safeguards]].

**Are these contradictory?** They operate at different altitudes and largely stack: board-level portfolio limits (A028) → cross-functional joint governance (A027) → decentralized LOB accountability (A058) → deregulated frontline access with targeted guardrails (A087). The honest synthesis — flagged in A058's own enrichment — is **centralized governance + decentralized accountability.** The unresolved edges are ethics standardization across decentralized owners ([[question-ethical-judgment-scale]]) and legal accountability under [[concept-accountability-blurring]] ([[question-legal-accountability]]). Closely tied to [[cross-colleague-or-tool]].