MAS Project MindForge — AI Risk Management: Operationalisation Handbook · v2024

MAS Project MindForge — AI Risk Management: Operationalisation Handbook

MAS Project MindForge — AI Risk Management: Operationalisation Handbook — 50% of in-scope requirements covered.

7 requirements · 7 in scope (7 designed). The 50% is weighted over the in-scope base.

Source: MAS Project MindForge is an industry-co-created supervisory initiative whose AI Risk Management: Operationalisation Handbook gives Singapore financial institutions practical guidance for operationalising AI risk management across four blocks — Scope & AI Oversight, AI Risk Management, AI Lifecycle Management, and Enablers (17 considerations in total). KYE Protocol operationalises the action-boundary subset of the Handbook at runtime: it governs whether a consequential financial AI action is authorised, within an oversight mode the FI has declared for that use, evidenced, contestable, and final at the moment it happens — and PROVES the basis, replayable by MAS or internal audit. KYE does NOT replace MindForge and does NOT govern the FI's governance operating model, model-development methodology, or the correctness of the AI's output (§0.25 integrate-not-compete; honest scope). · License: MAS Project MindForge — AI Risk Management: Operationalisation Handbook is published by the Monetary Authority of Singapore and industry participants. KYE registry cites the Handbook's oversight-mode and risk-management constructs for mapping purposes only.

By category

CategoryReqsEnforcedDesignedAdvisoryDeferredCoverage
Scope & AI oversight 2 0 2 0 0 50%
AI risk management 2 0 2 0 0 50%
AI lifecycle management 2 0 2 0 0 50%
Enablers 1 0 1 0 0 50%

Every requirement → the KYE artefact that enforces it

IDTitleStatusKYE enforcement
mas-mindforge.oversight-modes-machine-enforceable Scope & AI oversight: the three human-oversight modes are machine-enforceable at the action boundary, not policy-on-paper designed rule_packs: kye:rule-pack:mas-mindforge
dictionaries: internal
constitution_refs: constitution/36-GOVERNEDUI.md, constitution/12-PURPOSE-PERMISSION.md
mas-mindforge.action-authority-inventory Scope & AI oversight: the AI inventory is extended to an AI action-authority inventory — what each AI is AUTHORISED to do, not just what exists designed rule_packs: kye:rule-pack:mas-mindforge
dictionaries: internal
constitution_refs: constitution/12-PURPOSE-PERMISSION.md
mas-mindforge.escalation-before-finality AI risk management: out-of-mode and out-of-scope actions escalate (REQUIRE_APPROVAL) or are refused before finality designed rule_packs: kye:rule-pack:mas-mindforge
constitution_refs: constitution/25-EDGE-GOVERNANCE.md, constitution/36-GOVERNEDUI.md
mas-mindforge.third-party-vendor-authority-register AI risk management: third-party / vendor AI risk is governed via an external authority register at the action boundary designed rule_packs: kye:rule-pack:mas-mindforge
dictionaries: internal
constitution_refs: constitution/12-PURPOSE-PERMISSION.md
mas-mindforge.deployment-controls-authority-gateway AI lifecycle management: deployment controls are enforced at the Authority Gateway before a consequential action proceeds designed rule_packs: kye:rule-pack:mas-mindforge
constitution_refs: constitution/25-EDGE-GOVERNANCE.md, constitution/12-PURPOSE-PERMISSION.md
mas-mindforge.monitoring-change-evidence-replay AI lifecycle management: monitoring & change management produce replay-provable Evidence Packs, replayable by MAS or internal audit designed rule_packs: kye:rule-pack:mas-mindforge
dictionaries: internal
constitution_refs: constitution/13-RESILIENCE-LOOP.md, constitution/30-AUDIT-WORM-RETENTION.md
mas-mindforge.enablers-named-accountability Enablers: every consequential AI action maps to a named accountable officer under whose authority it proceeds designed rule_packs: kye:rule-pack:mas-mindforge
constitution_refs: constitution/36-GOVERNEDUI.md, constitution/21-DELEGATED-AUDITABILITY.md