MIT AI Risk Repository — Domain Taxonomy

MIT AI Risk Repository — Domain Taxonomy — 40% of in-scope requirements covered.

5 requirements · 5 in scope (3 designed · 2 advisory). The 40% is weighted over the in-scope base.

MIT AI Risk Repository: requirement → the KYE™ artefact that enforces it

Demo data

You learn: Which requirements KYE™ enforces at runtime and with which artefact. You can: Walk the requirements.

  1. d4-malicious-misuse · designed

    Domain 4 Malicious Actors & Misuse — unauthorised agent actions refused at the boundary

    Constitution: 12 purpose permission, 52 delegated agent binding

  2. d5-human-computer-interaction · designed

    Domain 5 Human-Computer Interaction — overreliance / loss of agency met with human oversight

    Constitution: 36 governedui

  3. d7-multi-agent · designed

    Domain 7 AI System Safety — multi-agent risks via the delegation chain

    Constitution: 52 delegated agent binding, 21 delegated auditability

  4. d2-privacy-security · advisory

    Domain 2 Privacy & Security — agent data access bounded by purpose + tenant isolation

    Constitution: 12 purpose permission, 31 data governance pack · Audit events: kye.evidence.decision_map.v1

  5. d7-traceability · advisory

    Domain 7 AI System Safety — traceability & non-repudiation of agent actions

    Constitution: 13 resilience loop · Audit events: kye.evidence.pack.v1, kye.replay.proof.v1

How this widget is built
Demonstrates
KYE™ Compliance Mapping Rail™
Components
  • Framework requirement (clause)
  • Coverage status
  • KYE™ enforcement artefact (event, decision-record field, Evidence Pack™ entry)
Flow
Clause → status → the KYE™ artefact that enforces it, enforced requirements first. Architecture diagram
Used on

Source: MIT AI Risk Repository (Slattery et al., 2024) — the Domain Taxonomy of 7 risk domains and 24 subdomains, the most-cited academic catalogue of AI risks. KYE Protocol™ governs the AUTHORITY, EVIDENCE and FINALITY of AI-AGENT ACTIONS; it therefore addresses the action-authority risk domains (privacy/security access, malicious misuse, human-computer oversight, system-safety traceability) and is HONESTLY out of scope for the content/societal domains (discrimination & toxicity, misinformation, socioeconomic & environmental) — those are content-truth/fairness/macro risks KYE™ does not adjudicate. Starter requirement set over the in-scope domains; deepen by graft through the §70 rail. · License: MIT AI Risk Repository is published under CC-BY; KYE™ registry paraphrases each domain's intent and cites the domain identifier for mapping purposes only.

By category

CategoryReqsEnforcedDesignedAdvisoryDeferredCoverage
Privacy & Security (action-authority)1001025%
Malicious Actors & Misuse1010050%
Human-Computer Interaction1010050%
AI System Safety, Failures & Limitations2011038%

Every requirement → the KYE™ artefact that enforces it

IDTitleStatusKYE™ enforcement
mit-risk.d2-privacy-securityDomain 2 Privacy & Security — agent data access bounded by purpose + tenant isolationadvisoryconstitution_refs: constitution/12-PURPOSE-PERMISSION.md, constitution/31-DATA-GOVERNANCE-PACK.md
audit_events: kye.evidence.decision_map.v1
mit-risk.d4-malicious-misuseDomain 4 Malicious Actors & Misuse — unauthorised agent actions refused at the boundarydesignedconstitution_refs: constitution/12-PURPOSE-PERMISSION.md, constitution/52-DELEGATED-AGENT-BINDING.md
mit-risk.d5-human-computer-interactionDomain 5 Human-Computer Interaction — overreliance / loss of agency met with human oversightdesignedconstitution_refs: constitution/36-GOVERNEDUI.md
mit-risk.d7-traceabilityDomain 7 AI System Safety — traceability & non-repudiation of agent actionsadvisoryconstitution_refs: constitution/13-RESILIENCE-LOOP.md
audit_events: kye.evidence.pack.v1, kye.replay.proof.v1
mit-risk.d7-multi-agentDomain 7 AI System Safety — multi-agent risks via the delegation chaindesignedconstitution_refs: constitution/52-DELEGATED-AGENT-BINDING.md, constitution/21-DELEGATED-AUDITABILITY.md

Canonical KYE™ surfaces referenced on this page: KYE Protocol™ · Systemic Authority Risk™.