Authority Envelope
The machine-enforceable expression of an authority grant — the runtime boundary beyond which an agent halts or escalates rather than proceeds.
- StatusProposed by QIS (not yet industry usage)
- AdoptionExperimental
- LayerGovernance
Why it matters
A grant stated in a policy document is a preference. An envelope is the same grant expressed where it can actually stop execution. The distinction matters because every published framework requires bounded autonomy and none specifies where the bound lives.
This term is a QIS Canonical Extension. It does not currently appear in the Model Context Protocol specification, the A2A (Agent2Agent) specification, the AIEOG Shared AI Lexicon, or any instrument in our standards register. We are proposing it because we have documented the gap it fills, and we label it as ours everywhere it appears. Adopt it, argue with it, or ignore it — but do not cite it as industry consensus, because it is not.
Where it comes from
QIS Canonical Extension. Names the enforced boundary that IMDA's framework and the FS AI RMF both require in prose without giving it a term.
Proposed here; instruments its origin note discusses
- Financial Services AI Risk Management Framework (FS AI RMF)Cyber Risk Institute · FSSCC (Financial Services Sector Coordinating Council) · US Treasury (AIEOG)Official text In the standards register
What it is not
These are routinely confused with Authority Envelope. The distinctions have consequences for how a system is governed.
- Authority Grant
The grant is the instrument. The envelope is its enforcement. A grant with no envelope is unenforced by definition.
How it connects
What relies on it
- Authority Envelope puts into practice Authority Grant Canon term
Where it hands decisions up
- Authority Envelope hands decisions up to Human Override Canon term
MARQUE questions it addresses
- AuthorityWhat is the agent permitted to commit, and who granted that?
Related framework objectives
Connected through the MARQUE (a reference framework organized around eight questions) crosswalk, an editorial mapping. These objectives from the Financial Services AI Risk Management Framework are assigned to the same questions that this term addresses. The ones sharing most of this term's vocabulary are shown first, as a reading aid. The assignment is to the question, not to the term: the source does not map this term to any objective directly.
Question: Authority · 39 control objectives assigned
- GV-2.1.1AI Risk Management Roles and ResponsibilitiesGovern
- GV-2.1.2AI Risk Communication and EscalationGovern
- MG-1.3.2Responsibility and Decision CriteriaManage
Cite and reuse
Cite the identifier QIS-TERM-00022, not the URL: identifiers are stable, URLs may change. This entry is free to read, quote and index under the dual license. Corrections sent to editor@qisfund.com are published.
Technical details: identifier, graph relationships and record
| Predicate | Target identifier | Target | Predicate meaning |
|---|---|---|---|
implements | term:authority-grant | Authority Grant | Read the other way round: the object is put into practice by the subject. The object is a concrete realization of the subject. |
addresses | marque:authority | Authority | The subject speaks to the object as a question or concern. |
escalatesTo | term:human-override | Human Override | The subject hands a decision upward to the object. |
relatedTo | cat:constraints | Constraints | An association too weak or too general for a stronger verb. |
relatedTo | term:kill-functionality | Kill Functionality | An association too weak or too general for a stronger verb. |
- Canonical identifier
QIS-TERM-00022- Tier
- QIS Canonical Extension
- Domain · layer
- Authority · Governance
- Semantic aliases
- None recorded
- First published
- 2026-08-02
- Last reviewed
- 2026-08-02 · 180-day cycle