Singapore's Monetary Authority of Singapore (MAS) announced plans to develop a standardized protocol for AI agents to conduct financial transactions, with a dedicated sandbox environment for testing payments using stablecoins and tokenized deposits [1]. Chia Der Jiun, managing director of MAS, outlined the proposal on November 19, 2024, at the Singapore FinTech Festival, introducing what the central bank calls the "know-your-agent protocol" [1]. This protocol would enable financial institutions to verify an AI agent's identity, the customer it represents, its delegated authority, and the rules it must follow, while also conveying assurances about the agent's origin and reliability [1].
The proposed framework includes two main trial initiatives under the Agentic Money Sandbox, announced jointly by MAS and the Future of Finance Institute on November 18 [1]. The first trial would test direct payments, where a customer initiates a payment, an agent submits the instruction, and the bank or payment provider verifies the agent's identity and authority before executing the transaction using digital money such as tokenized deposits or stablecoins [1]. The second trial would focus on corporate treasury management, allowing companies to define what their agents can do—including monitoring cash balances, deciding when and where funds should move, and executing transfers using tokenized deposits—while banks set their own controls and terms for accepting instructions [1].
The governance framework establishes clear responsibility boundaries: customers define what agents can do and under what limits, AI agents send instructions on the customer's behalf, and financial institutions verify identity and authority while applying risk controls [1]. According to MAS's approach, the strength of controls should scale with the agent's discretion and potential impact [1]. This initiative represents a significant step toward operationalizing AI agents in financial services, with MAS planning to help develop the protocol further through its Safeguards for Agentic Finance at Runtime initiative [1].
CONCLUSION
Singapore's MAS is establishing a regulatory framework and testing environment for AI agents to conduct financial transactions using stablecoins and tokenized deposits, positioning the city-state as a leader in agentic finance governance. The proposed know-your-agent protocol and dual-track sandbox trials aim to balance innovation with risk management, creating standardized verification and control mechanisms for AI-driven payments and treasury operations.