The Project MindForge AI Risk Management Toolkit was announced by MAS on 20 March 2026 as the outcome of phase two of Project MindForge, developed with 24 financial institutions. Its central component is the AI Risk Management Operationalisation Handbook (cover dated January 2026), which sets out a MindForge AI Risk Management and Governance Framework in four parts: scope and AI oversight, AI risk management, AI lifecycle management and enablers. It is an industry-led, voluntary companion to MAS's proposed Guidelines on AI Risk Management rather than a MAS requirement. Phase one of MindForge had produced a generative AI risk framework in November 2023.
OFFICIAL TEXT: mas.gov.sg ↗ · FINAL · MAS
| Document | MAS MindForge AI Risk Management Toolkit — AI Risk Management: Operationalisation Handbook (Project MindForge AI Risk Management Toolkit) |
| Issued by | Monetary Authority of Singapore |
| Type | Framework |
| Status | Final |
| Published | Mar 20, 2026 |
| Applies to | Financial institutions in Singapore and elsewhere that want practical guidance on AI governance and risk management; developed by a 24-institution industry consortium (banks, insurers and capital markets firms) with MAS support. It is an industry-led companion guide, not a MAS rule or a binding standard |
| Official source | mas.gov.sg ↗ |
| Use cases | AI governance (general) · Model risk management · Generative & agentic AI · Third-party & vendor AI |
What are the key points of MAS MindForge AI Risk Management Toolkit?
- The toolkit comprises the Operationalisation Handbook and an AI case-studies supplement according to the MAS media release; the MindForge page also lists an Executive Handbook and an implementation-examples document.
- The Handbook's framework has four sections: (1) scope and AI oversight, (2) AI risk management, (3) AI lifecycle management and (4) enablers, structured around 17 numbered Considerations.
- Section 2 covers AI policies, procedures and standards, organisation-level risk management, third-party AI risks, use-case-level risk management and AI inventory capabilities.
- Section 3 follows the AI lifecycle: use case context and design, data acquisition and processing, onboarding, build and review, deployment, and usage, monitoring and change management.
- Section 4 covers skills, knowledge and culture, and AI infrastructure; appendices include a glossary, a MindForge AI Risk Taxonomy, an AI Card template, libraries of AI metrics and guardrails, and an AI risk management checklist.
- Part 3 (Future Perspectives) compares generative AI and agentic AI architectures.
- The November 2025 MAS consultation described the Handbook as an industry-led companion guide for FIs implementing the proposed Guidelines.
- MAS said in the release that it will set up an AI risk management workgroup under BuildFin.ai to develop implementation resources and address newer technologies such as agentic AI.
What did MAS MindForge AI Risk Management Toolkit change for banks?
MindForge phase one (whitepaper 'Emerging Risks and Opportunities of Generative AI for Banks', November 2023) identified generative AI risks. Phase two turns the proposed MAS Guidelines into an operational playbook with control libraries, templates and case studies, extending the consortium from banks to insurers and capital markets firms.
What does the MindForge AI Risk Management Operationalisation Handbook recommend for banks?
The MindForge AI Risk Management Operationalisation Handbook, part of the toolkit MAS announced on 20 March 2026, sets out a MindForge AI Risk Management and Governance Framework with four sections: scope and AI oversight; AI risk management (policies, organisation-level and use-case-level risk management, third-party AI risks and AI inventories); AI lifecycle management (design, data, build and review, deployment, monitoring and change); and enablers (skills, culture and infrastructure). It supplies 17 Considerations with practical examples, plus templates, libraries of AI metrics and guardrails, a risk taxonomy and a checklist. It is industry-led, voluntary and designed as a companion to MAS's proposed Guidelines on AI Risk Management, not a regulation.
| Rule | Authority | What it requires | Applies |
|---|---|---|---|
| Section 1 — Scope and AI oversight | MAS | Define the scope and application of AI governance and assign responsibilities for AI oversight (Consideration 1). | Voluntary guidance, March 2026 |
| Section 2.1-2.2 — Policies and organisation-level risk | MAS | Maintain effective AI policies, procedures and standards and enhance organisation-level AI risk management (Considerations 2-3). | Voluntary guidance, March 2026 |
| Section 2.3 — Third-party AI risks | MAS | Uplift practices for managing third-party AI risks, including pre- and post-procurement mitigations (Consideration 4). | Voluntary guidance, March 2026 |
| Section 2.4-2.5 — Use-case risk and inventory | MAS | Run use-case-level AI risk management with materiality assessment, and maintain an AI inventory (Considerations 5-6). | Voluntary guidance, March 2026 |
| Section 3 — AI lifecycle management | MAS | Apply controls across design, data acquisition, build and review, deployment, and usage, monitoring and change management (Considerations 7-15). | Voluntary guidance, March 2026 |
| Section 4 — Enablers | MAS | Build AI governance skills, knowledge and culture, and manage AI infrastructure (Considerations 16-17). | Voluntary guidance, March 2026 |
The Handbook adopts the same terms as MAS's proposed Guidelines (AI model, system and use case, and the AI lifecycle) and places FEAT, Veritas and MindForge in one picture of the Singapore stack. MAS's March 2026 release said existing toolkits will be updated, and a BuildFin.ai workgroup will address newer technologies such as agentic AI.
Because it is industry-led, it is best used as an implementation reference. The supervisory yardstick remains the MAS Guidelines once finalised, and the December 2024 information paper for observed practice.
WHAT THIS MEANS IN PRACTICE
- Use the 17 Considerations as a gap-assessment checklist against your AI governance framework.
- Adopt the AI Card template and control library concepts for inventory and use-case review.
- Use the MindForge AI Risk Taxonomy to harmonise risk language across business lines.
- Track MAS's final AI Risk Management Guidelines and the BuildFin.ai workgroup outputs for changes.
- Do not cite the Handbook to supervisors as a MAS requirement; it is voluntary industry guidance.
Does the MindForge AI Risk Management Toolkit apply to banks?
It is written for financial institutions generally, including banks, and was developed by a 24-institution consortium of banks, insurers and capital markets firms. It is voluntary guidance and does not create binding obligations.
Is the MindForge Handbook a MAS requirement?
No. MAS's consultation paper describes it as an industry-led handbook serving as a companion guide for FIs implementing the proposed Guidelines on AI Risk Management. The binding or supervisory expectations remain those MAS issues itself.
What was the earlier MindForge generative AI paper?
Phase one of MindForge, led by a bank consortium, released a risk framework for generative AI in November 2023, published as the whitepaper 'Emerging Risks and Opportunities of Generative AI for Banks'; its executive summary was released on 15 November 2023.
When was the MindForge toolkit published?
MAS announced the conclusion of phase two and the toolkit on 20 March 2026. The Operationalisation Handbook's cover is dated January 2026, and the November 2025 consultation said it would be released by January 2026.
| Date | Document | Status |
|---|---|---|
| Nov 13, 2025 | MAS Consultation Paper P017-2025 (AI Risk Management Guidelines) — Consultation Paper on Guidelines on Artificial Intelligence Risk Management | Proposed |
| Sep 18, 2025 | MAS Circular MAS/TCRS/2025/06 — Cyber Risks Associated with Deepfakes (Information Paper) | In force |
| Dec 5, 2024 | MAS AI Model Risk Management information paper — Artificial Intelligence Model Risk Management: Observations from a Thematic Review (Information Paper) | Final |
| Jul 30, 2024 | MAS Circular MAS/TCRS/2024/05 — Cyber Risks Associated with Generative Artificial Intelligence (Information Paper) | In force |
| Nov 12, 2018 | MAS FEAT Principles — Principles to Promote Fairness, Ethics, Accountability and Transparency (FEAT) in the Use of Artificial Intelligence and Data Analytics in Singapore's Financial Sector | In force |
| Sep 30, 2026 | SB 947 — Employment: Automated Decision Systems (No Robo Bosses Act) | Final |
Follow every document these regulators publish
when one of these regulators moves, the next morning's brief says so · six sourced stories · 7 am ET · free
plus every tracker, bank and agent page update, the morning after · leave any morning