MAS Sets Out Safeguards for Autonomous AI Agents in Finance

The Monetary Authority of Singapore has released a white paper, developed with banks and fintech firms, setting out safeguards for the use of autonomous artificial intelligence agents in financial services. The paper addresses a fast-emerging challenge: as AI agents increasingly execute tasks such as processing transactions, monitoring risk and interacting with customers, traditional human oversight of every action becomes impractical at scale. To manage the risks, the regulator proposes runtime safeguards, controls that operate while the AI is running, including mechanisms to constrain agent behaviour, monitor actions in real time, enforce limits and intervene or halt processes when something goes wrong. The framework aims to let financial institutions harness the efficiency and speed of agentic AI while maintaining accountability, security and consumer protection. It builds on Singapore's broader work on AI governance, model risk management and operational resilience, reinforcing the city-state's position as a hub for responsible financial innovation. By convening industry to co-develop practical standards, MAS is seeking to stay ahead of a technology that is advancing rapidly, ensuring safeguards keep pace with deployment. The guidance is likely to inform how banks and insurers across the region approach the governance of increasingly autonomous systems.