Singapore's Central Bank to Create Standard Framework for AI Payment Agents
Singapore's central bank unveils plans for AI agent payment tests using stablecoins, enhancing identity verification and risk control protocols.

Introduction
In a significant move to advance the integration of artificial intelligence in financial services, Singapore's central bank is set to establish a standardized framework for banks to validate the identities of AI agents and the authorizations granted by customers. This initiative was announced by Chia Der Jiun, the managing director of the Monetary Authority of Singapore (MAS), during the Singapore FinTech Festival on November 19.
Know-Your-Agent Protocol
The proposed framework, termed the know-your-agent protocol, would provide crucial information to financial institutions about the identity of an AI agent, its designated customer, the extent of its authorization, and the parameters within which it must operate. This measure aims to ensure that banks can accurately assess the trustworthiness and reliability of AI agents, thereby enhancing security for digital transactions.
Agentic Money Sandbox
Additionally, MAS announced the introduction of the Agentic Money Sandbox, in collaboration with the Future of Finance Institute, which will create a controlled environment where firms can safely experiment with payment systems utilizing digital currencies, including stablecoins and tokenized deposits. This sandbox will allow financial institutions to innovate while managing the associated risks more effectively.
Testing Environment and Customer Mandate
In the proposed testing environment, a participating bank or payment provider will take steps to confirm the agent's identity and the authority it possesses before proceeding with transactions. The extent and authority of the agent's actions will be outlined in a customer mandate, which is vital for defining the boundaries of what the AI agent may do—especially in scenarios where the agent’s discretion could significantly impact funds management.
Control and Oversight
Chia emphasized that the level of control applied should correlate with the degree of discretion afforded to the agent, suggesting that more significant responsibilities would necessitate stricter oversight.
Accountability and Future Implications
During a prior speech on November 18, Singapore's Deputy Prime Minister and MAS Chairman Gan Kim Yong illustrated a scenario where an AI agent could process payments once it verifies that the contract conditions have been satisfied.
Notably, the launch partners for this initiative include major banks such as DBS, OCBC, Standard Chartered, and UOB. However, details regarding the timeline for the sandbox’s implementation and a comprehensive list of its participants have yet to be disclosed.
Gan also stressed that it remains crucial for institutions and companies to maintain accountability for the actions of their agents, raising pertinent questions about liability in instances where an AI agent acts beyond its mandates or makes errors. With these developments, MAS is moving toward a future where AI can play a pivotal role in financial operations while ensuring robust safeguards are in place.
Also in other languages: РусскийDeutschFrançaisItalianoEspañol