The Pulse
Singapore’s MAS Sets AI Risk Expectations for Financial Firms
Singapore’s Monetary Authority has issued AI risk-management guidelines covering all financial institutions and AI technologies. The framework sets staged implementation dates and makes firms accountable for AI supplied by third parties.
AI.info Team · Written with AI ·
MAS applies the framework across Singapore’s financial sector
Singapore’s Monetary Authority (MAS) issued AI risk-management guidelines on 7 October 2026, setting supervisory expectations for financial institutions as they adopt AI. The guidance applies to all financial institutions and all forms of AI technology, from generative AI to systems that can act with greater autonomy. MAS says institutions should manage risks both across the organisation and for individual use cases.
The guidelines are designed to scale with each institution’s activities and exposure, rather than impose identical controls on every use of AI. They cover oversight, AI inventories and risk assessments, controls throughout the technology’s life cycle, and the skills and resources institutions need to manage it. The guidance complements existing MAS requirements and principles, including Singapore’s FEAT principles on fairness, ethics, accountability and transparency.
Boards remain accountable without a new AI committee
MAS expects boards and senior management to set clear responsibilities, define risk appetite and ensure that existing risk-management systems account for AI. Institutions can use their current governance structures if those arrangements provide effective oversight and coordination across relevant teams. The guidance does not require a dedicated AI committee solely to meet its expectations.
“AI has significant potential to improve financial services, from enhancing customer outcomes and strengthening risk management to improving productivity and enabling new products and services,” said Ho Hern Shin, MAS deputy managing director. She said institutions need to understand and manage risks as AI systems become more capable, so they can adopt the technology while maintaining customer trust and the resilience of Singapore’s financial system.
Firms must track use cases and assess third-party tools
The guidelines call for institutions to identify their AI use, maintain inventories at an appropriate level of detail and assess the materiality of risks for each use case. Depending on the application, controls may cover data governance, testing, human oversight, cybersecurity, monitoring and change management. MAS says institutions should review those controls as their AI use expands and technology changes.
Responsibility also extends to AI embedded in services supplied by outside firms. Financial institutions remain accountable for AI used in services they provide, including systems developed or operated by third parties. MAS expects them to assess whether those systems suit their intended purpose and obtain sufficient assurance from providers; where gaps remain, they should use compensating controls where practical. If a provider’s risks cannot be brought within the institution’s risk appetite, MAS says it should consider limiting, suspending or replacing the service.
Deadlines stretch from 2027 to 2028
The guidelines take effect on 7 October 2027. Institutions should meet the expectations in Sections 3 and 4—covering oversight and risk-management systems—by that date, and meet the life-cycle control and capability expectations in Sections 5 and 6 by 7 October 2028. MAS allows phased implementation, while expecting firms to calibrate the detail of their controls to their risk exposure.
Lower-impact uses may receive simpler governance. The guidelines give examples such as AI-assisted email drafting or internal document summaries, provided poor performance or an outage is unlikely to materially affect the institution, its customers or other stakeholders. MAS also plans to consult the financial sector in 2027 on whether further guidance is needed for agentic AI systems, which can operate autonomously and access tools. The full guidelines set out the expectations and phased deadlines.