Singapore's MAS Backs Industry Framework to Keep AI Agents in Finance Safe

The Monetary Authority of Singapore has teamed up with major financial institutions and fintech firms to release a framework designed to keep artificial intelligence agents in finance operating safely as they take on more autonomous tasks. The white paper, called Safeguards for Agentic Finance at Runtime, sets out an industry-built approach to ensuring AI agents perform financial activities in a way that is secure, dependable and within clearly defined boundaries. It was produced under a MAS-backed programme that supports the responsible development and deployment of AI across the financial sector. The initiative responds to the rapid rise of so-called agentic AI, in which software can plan and execute multi-step tasks with limited human oversight, at a pace that can make real-time human monitoring impractical. To manage that, the framework emphasises runtime controls that keep agent behaviour inside the mandates, policies and risk limits that firms set, rather than relying only on checks before deployment. For banks and insurers, the guidance points toward guardrails covering authorisation, monitoring and the ability to intervene or halt agents. The effort reflects Singapore's broader strategy of encouraging financial innovation while building safeguards to protect consumers and the stability of the system.