The Monetary Authority of Singapore concluded the second phase of Project MindForge on 20 March 2026, publishing an AI Risk Management Toolkit developed with a consortium of 24 banks, insurers, capital market firms, and industry partners. The toolkit represents the most comprehensive AI governance framework produced for the financial services sector in Southeast Asia to date, covering traditional AI, generative AI, and emerging agentic AI technologies.
Key Facts At A Glance
- MAS announced the conclusion of Project MindForge Phase Two on 20 March 2026.
- The AI Risk Management Toolkit was developed by a consortium of 24 institutions across banking, insurance, and capital markets.
- Phase Two expanded the scope beyond banking to include insurance and asset management sectors.
- The toolkit includes an AI Risk Management Operationalisation Handbook structured around four areas: scope and oversight, AI risk management, AI lifecycle management, and organisational enablers.
- A supplementary case study compilation documents implementation experiences and lessons from participating institutions.
- Consortium members include Citi, DBS, GXS, HSBC, Julius Baer, Maybank, MUFG Bank, OCBC, SMBC, Standard Chartered Bank, UBS, UOB, AIA, Great Eastern Life, HSBC Life, Income Insurance, Manulife, MSIG, Munich Re, Prudential, BlackRock, Eastspring Investments, GIC, and State Street.
- Technology and consulting partners include Accenture, AWS, Google Cloud, Microsoft, and Nvidia.
- MAS is still reviewing responses from its earlier public consultation on proposed Guidelines on AI Risk Management.
- A new AI risk management workgroup will be established under MAS’s BuildFin.ai initiative to support ongoing implementation.
Project MindForge was launched by MAS in mid-2023 to strengthen AI risk management for financial institutions deploying AI in their services and operations. Phase One, concluded in November 2023, produced a generative AI risk framework and a platform-agnostic reference architecture. Phase Two expanded the scope significantly, broadening both the institutional participation and the range of AI technologies covered to include traditional AI, generative AI, and agentic AI.
Toolkit Structure And Content
At the core of the Phase Two output is the AI Risk Management Operationalisation Handbook, which provides practical guidance on how financial institutions can implement AI risk controls within their existing operating models. The handbook is organised into four sections aligned with MAS’s proposed Guidelines on AI Risk Management: scope and oversight, AI risk management, AI lifecycle management, and the capabilities and resources needed to support responsible AI use. A supplementary section compiles case studies from participating institutions, documenting implementation challenges, applied approaches, and lessons learned across different AI deployment environments.
MAS noted that the handbook will be periodically updated to reflect the maturation of AI use in the industry and to align with evolving supervisory expectations. The regulator said it is continuing to review feedback from its earlier public consultation on the proposed guidelines, and no final version of those guidelines has been published as of the date of this report.
Consortium Composition
The Phase Two consortium drew participation from across the financial sector. Banks involved include Citi, DBS, GXS, HSBC, Julius Baer, Maybank, MUFG Bank, OCBC, SMBC, Standard Chartered Bank, UBS, and UOB. Insurers include AIA, Great Eastern Life, HSBC Life, Income Insurance, Manulife, MSIG, Munich Re, and Prudential. Capital market participants include BlackRock, Eastspring Investments, GIC, and State Street. Industry associations involved are the Association of Banks in Singapore, the General Insurance Association, the Investment Management Association of Singapore, the Life Insurance Association, and the Singapore Fintech Association. Technology and consulting partners include Accenture, AWS, Google Cloud, Microsoft, and Nvidia.
BuildFin.ai And Next Steps
To support broader industry adoption, MAS will establish a dedicated AI risk management workgroup under its BuildFin.ai initiative, which brings together technology providers, research institutes, and financial institutions to develop shared solutions for common industry challenges. The workgroup will focus on implementation resources and governance frameworks for newer AI technologies, particularly agentic AI systems that can set objectives, organise activities, and operate with reduced human oversight.
Kenneth Gay, Chief FinTech Officer at MAS, said the toolkit marked a significant step forward in promoting responsible AI adoption in finance and that the BuildFin.ai programme would serve as the foundation for the next phase of AI risk management collaboration across the industry.
Regulatory Context
The MindForge toolkit follows a 2024 MAS information paper on good practices in AI and generative AI model risk management, and a 2025 partnership between MAS and the UK’s Financial Conduct Authority on AI in financial services. The toolkit’s structure mirrors the four-pillar framework in MAS’s proposed AI risk management guidelines, giving financial institutions a practical instrument for preparing for formal supervisory requirements once those guidelines are finalised.