Guardrails
Layered safeguards — policies, technical controls and monitoring — applied at the data, model, application and infrastructure levels to keep a system operating within organizational or regulatory boundaries.
- StatusIndustry term
- AdoptionWidely used
- LayerGovernance
Why it matters
Guardrails shape behavior continuously. They are not a stopping mechanism, and presenting them as one is how institutions end up with no stopping mechanism.
Where it comes from
Defined in the AIEOG Shared AI Lexicon (February 2026).
Defining sources
- AIEOG Shared AI LexiconUS Treasury · FBIIC (Financial and Banking Information Infrastructure Committee) · FSSCC (Financial Services Sector Coordinating Council)Official text In the standards register
What it is not
These are routinely confused with Guardrails. The distinctions have consequences for how a system is governed.
- Kill Criteria
Guardrails constrain how a system operates. Kill criteria determine whether it continues to operate at all.
- Authority Grant
Guardrails are applied by the operator to the system. A grant is issued by a principal and defines what the system may commit.
How it connects
What relies on it
- Guardrails sets limits on Agent Identity Canon term
MARQUE questions it addresses
- OversightWho is accountable in the interval between decision and review?
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: Oversight · 110 control objectives assigned
- GV-1.2.5AI Governance and Risk StandardsGovern
- MG-3.1.2Vendor Risk and Compliance AssessmentsManage
- GV-1.4.2AI Risk Documentation StandardsGovern
Cite and reuse
Cite the identifier QIS-TERM-00015, 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 |
|---|---|---|---|
constrains | term:agent-identity | Agent Identity | The subject bounds what the object may do. |
addresses | marque:oversight | Oversight | The subject speaks to the object as a question or concern. |
- Canonical identifier
QIS-TERM-00015- Tier
- Canonical industry term
- Domain · layer
- Oversight · Governance
- Semantic aliases
- None recorded
- First published
- 2026-08-02
- Last reviewed
- 2026-08-02 · 180-day cycle