The Monetary Authority of Singapore (MAS) finalised its AI Risk Management Guidelines on 7 October 2026, setting supervisory expectations for banks and other regulated financial institutions using predictive, generative and agentic AI. The framework moves beyond responsible-AI principles by requiring board and senior-management oversight, an active inventory of internally developed, third-party and embedded AI, and risk assessments based on impact, complexity and reliance on human intervention. Governance, identification and materiality requirements take effect on 7 October 2027; lifecycle controls and staffing requirements follow on 7 October 2028. Higher-risk uses call for independent pre-deployment validation, data governance, bias and cyber testing, ongoing monitoring and workable human overrides or fallback processes. Institutions remain accountable for vendor systems and should seek independent assurance rather than rely solely on suppliers’ claims. For banks, the timetable makes AI visibility and evidence of controls pressing priorities, while increasing demands on technology providers for assurance. MAS plans a separate consultation on agentic AI in 2027.