Govern (GV) · Authorization

Authorization at the Organizational charter layer

external · Outside the agent entirely

Who is allowed to put this agent into the world, under what authority, accountable to what policy, with what retirement criteria?

What this cell does

AI Risk Appetite Statement defines hard prohibitions and change-control process for evolving scope policy.

Artifacts (1)

checks.yamlview on GitHub

# ABOUTME: Machine-checkable definition of the Charter authorization / organizational cell.
# ABOUTME: The audit prompts in this cell's README, expressed so a validator can score them.

cell:
  id: authorization.server-side
  concern: authorization
  layer: server-side
  authority: organizational
  document: organizational-policy
  owner: "AI Governance Council."
  question: "Does the org have an AI Risk Appetite Statement that names hard prohibitions every agent must obey, and a change-control process for evolving scope policy?"

mappings:
  csf: "GV.RM-01, GV.RM-02, GV.PO-02"
  ai_rmf: "GOVERN 1.3, MANAGE 1.1"
  iso42001: "A.5"
  eu_ai_act: "Art. 9"

checks:
  - id: GV-AZ-O-01
    description: "Hard prohibitions binding on every agent are enumerated."
    type: min_items
    target: "risk_appetite.hard_prohibitions:1"
    document: organizational-policy
    severity: blocking
    evidence: "Absolute prohibitions, not guidance."
  - id: GV-AZ-O-02
    description: "Risk appetite carries at least one quantitative cap."
    type: min_items
    target: "risk_appetite.quantitative_caps:1"
    document: organizational-policy
    severity: blocking
    evidence: "A number with a unit. 'Low risk' fails."
  - id: GV-AZ-O-03
    description: "A change-control process for scope policy is named."
    type: required_field
    target: "risk_appetite.change_control"
    document: organizational-policy
    severity: blocking
    evidence: "Who may amend, and through what process."

Cell notes

Charter, Authorization / Organizational

Structural question. Does the org have an AI Risk Appetite Statement that names hard prohibitions every agent must obey, and a change-control process for evolving scope policy?

Owner. AI Governance Council.

Template fragment

§2.2 (Hard prohibitions), §2.3 (Agents requiring elevated approval), and §3 (Risk Appetite) of ../../templates/organizational-policy.md.

Audit prompts

- What are the org's hard prohibitions for agents? Are they enforced at runtime via Covenants L3-C2?
- How does the org change scope policy? PR-with-Council-approval, or hand-edit?
- When was the policy last amended? Was the amendment recorded in Council minutes?

Operational tie-in

Hard prohibitions are codified in the Kyverno policies under ../../../controls/authorization/server-side/. If your AUP says "no agent has direct prod-database write," the corresponding agents-no-cluster-roles policy must include a deny rule for that specific resource.

Citation

NIST CSF 2.0 GV.RM-01, GV.RM-02 (risk management objectives, risk appetite); GV.PO-01. NIST AI RMF GOVERN 1.2 (transparency), GOVERN 1.5 (ongoing monitoring). ISO/IEC 42001 §A.6.1 (risk management). EU AI Act Art. 9(2) (risk management as continuous iterative process).

Crosswalk

NIST CSF 2 0	GV.RM-01, GV.RM-02, GV.PO-01
NIST AI RMF	GOVERN 1.2, GOVERN 1.5
ISO IEC 42001	§A.6.1
EU AI ACT	Art. 9(2)

Cite this cell: https://agenticcovenants.com/govern/authorization/server-side/