Singapore's Monetary Authority (MAS) has unveiled a comprehensive AI Risk Management Toolkit, marking a pivotal shift from broad principles to actionable controls for financial institutions navigating the risks of artificial intelligence. Developed in collaboration with 24 banks, insurers, capital markets firms, and industry partners through phase two of Project MindForge, the toolkit equips firms to manage threats across traditional AI, generative AI, and emerging agentic AI systems.

The release, announced on March 23, 2026, arrives as AI adoption accelerates in Asia's financial hub, where institutions are increasingly integrating advanced models for everything from fraud detection to customer service. At its core is the AI Risk Management Operationalisation Handbook, a practical guide structured around four key pillars that align with MAS's proposed Guidelines on AI Risk Management: scope and oversight, AI risk management, AI lifecycle management, and organisational enablers like capabilities, infrastructure, and resources.

FROM PRINCIPLES TO PRACTICE

Project MindForge, launched in mid-2023, represents MAS's broader agenda to foster responsible AI in finance. This initiative builds on a 2024 information paper outlining good practices for AI and generative AI model risk management, as well as a 2025 partnership with the UK's Financial Conduct Authority. The toolkit concludes phase two by translating supervisory expectations into operational reality, addressing gaps where many banks possess AI policies on paper but struggle with implementation for newer technologies.

The handbook provides step-by-step guidance on establishing governance frameworks, conducting risk materiality assessments, inventorising AI usage, and implementing lifecycle controls from development to deployment and monitoring. Accompanying it is a supplement of case studies from participating institutions, detailing real-world lessons, challenges, and tailored approaches in diverse organisational contexts. These examples highlight issues like model behaviour unpredictability in generative AI and accountability in agentic systems, which autonomously make decisions.

MAS Chief FinTech Officer Kenneth Gay emphasised the toolkit's significance, stating, “The development of the MindForge AI Risk Management Toolkit, including the release of the Operationalisation Handbook, marks a major step forward in our journey to ensure the responsible adoption of AI in finance. We are committed to fostering a culture of continuous engagement and strengthening of AI governance and risk management practices across the industry.”

COLLABORATIVE SUPERVISORY APPROACH

This industry-led development signals MAS's preference for collaboration over top-down regulation. Rather than issuing standalone rules, the authority partnered with heavyweights in Singapore's financial ecosystem, reflecting a maturing supervisory stance. The toolkit will evolve, with periodic updates to match advancing AI use and refined expectations following MAS's ongoing review of feedback from a November 2025 public consultation on the proposed guidelines.

To drive adoption, MAS plans an AI risk management workgroup under its BuildFin.ai initiative, uniting MindForge consortium members, technology providers, research institutes, and practitioners. This group will create further resources, facilitate knowledge sharing, and develop frameworks for agentic AI, which introduces novel risks around oversight and autonomy. BuildFin.ai already tackles shared challenges, positioning Singapore as a leader in scalable fintech innovation.

AI'S RISING STAKES IN FINANCE

Financial firms in Singapore and beyond face mounting pressures from AI's dual-edged potential. Generative AI, powering tools like chatbots and content creation, has proliferated since tools like ChatGPT emerged, while agentic AI—capable of executing complex tasks independently—promises efficiency but amplifies risks such as bias, data privacy breaches, and operational failures. Regulators globally are responding: the EU's AI Act classifies financial AI as high-risk, and the US Federal Reserve stresses model risk management.

In Asia, where AI budgets remain constrained amid thin margins, institutions are chasing competitive edges in areas like personalised banking and risk modelling. Financial institutions are more concerned about how artificial intelligence can enhance their market position. Almost 2 in 3 FIs allocate 10% or less of their tech budgets on AI. The MindForge toolkit directly counters gaps in governance, offering templates for inventorisation systems and control matrices that integrate with existing compliance processes.

Case studies in the supplement reveal practical hurdles faced by participating institutions, underscoring the toolkit's value in demystifying implementation, potentially reducing compliance costs while enhancing resilience.

GLOBAL IMPLICATIONS FOR HUBS

Singapore's move reverberates across global finance, where AI governance fragments persist. As a nexus for financial institutions, MAS's toolkit could become a benchmark, much like its green finance taxonomies. Industry observers praise the collaborative model, noting it transforms risk management from a burden to a strategic asset, enabling faster AI scaling.

Yet challenges remain. Agentic AI's opacity demands ongoing innovation, and cross-border harmonisation is elusive. MAS's workgroup may pioneer solutions, but firms must invest in talent and infrastructure. Gay's vision of “continuous engagement” hints at phase three, possibly incorporating standards for AI explainability and adversarial robustness.

For Singapore's banks and insurers, the toolkit is not mere guidance—it's a roadmap to responsibly harness AI's transformative power, safeguarding stability while fuelling growth in an era of intelligent automation.