Singapore's MAS Unveils Framework to Keep Autonomous AI Agents Safe in Finance

The Monetary Authority of Singapore has joined forces with major financial institutions and fintech firms to publish a white paper aimed at keeping autonomous artificial-intelligence agents in finance operating within safe limits. The framework, called Safeguards for Agentic Finance at Runtime, or SAFR, sets out an industry-built approach designed to let AI agents carry out financial tasks in a manner that is safe, secure and dependable. The initiative responds to the rapid rise of 'agentic' AI, in which software agents can plan and execute multi-step actions with limited human intervention, a shift that promises efficiency gains but also makes continuous human oversight increasingly impractical at scale. The paper focuses on runtime controls, meaning safeguards that operate while agents are actively performing tasks, to help ensure they stay within defined risk limits and behave predictably. It reflects Singapore's broader strategy of engaging industry directly to translate emerging technology into practical, well-governed tools, complementing the regulator's wider work on AI governance, operational resilience and financial innovation. By setting shared expectations early, the authority aims to support responsible adoption of autonomous systems while protecting consumers and the stability of the financial system.