The Monetary Authority of Singapore (MAS), in collaboration with leading financial institutions and FinTech companies, has released a white paper proposing a framework for safeguarding AI agents in the financial sector. The paper, titled “Safeguards for Agentic Finance at Runtime” (SAFR), was published on 3 July 2026 and is part of MAS’ BuildFin.ai initiative, which promotes responsible AI development in finance.
As AI agents increasingly perform tasks autonomously, the need for real-time safeguards is crucial to ensure their actions remain within set mandates and risk boundaries. The SAFR framework introduces governance checkpoints to verify and record AI actions before execution, ensuring compliance with predefined policies.
The framework builds on MAS’ Project Mindforge’s AI Risk Management toolkit, focusing on operationalising safeguards at the point of action. It outlines how policy-bound execution, real-time validation, auditability, and interoperability can be integrated into system operations, allowing financial institutions to deploy AI agents with greater trust.
Industry members have already applied the SAFR framework in various use cases, such as agent-assisted payments, wealth management, and client engagement. These applications demonstrate improved efficiency, reduced operational friction, and enhanced compliance and client interaction.
MAS invites interested industry partners to join the BuildFin.ai work group to further develop SAFR. The Future of Finance Institute will support the framework’s adoption through industry pilots and sandbox experimentation, facilitating the testing and deployment of SAFR-aligned solutions.
