MAS and Industry Publish Safeguards Framework for AI Agents in Finance

The Monetary Authority of Singapore has published an industry white paper on safeguards for artificial intelligence agents in finance, developed jointly with leading financial institutions and fintech companies. Titled Safeguards for Agentic Finance at Runtime, or SAFR, the paper proposes an industry-developed framework designed to allow AI agents to carry out financial tasks safely, securely and reliably. The distinction matters: unlike conventional AI tools that generate recommendations for human review, agentic systems can take actions autonomously, executing transactions or initiating processes without a person approving each step. That capability raises the stakes for controls around authorisation, auditability, error handling and the ability to intervene when an agent behaves unexpectedly. SAFR was developed under the regulator's BuildFin.ai initiative, a collaborative programme bringing together financial institutions, technology providers and research institutes to co-develop AI solutions addressing real market needs. The publication forms part of a wider set of moves from the authority, which has also established a Future of Finance Institute focused on artificial intelligence and tokenisation, consulted on a protected cell company framework for alternative risk transfer in insurance, and proposed changes to speed up approvals for new fund types.