Banks need to rethink identity governance for AI agents as artificial intelligence moves from assisting to acting autonomously. The Monetary Authority of Singapore published Safeguards for Agentic Finance at Runtime last July, which made explicit the need for AI agents to be governed like users with raised access.
The guidance is industry guidance, not a supervisory requirement, but its core recommendation is concrete. It calls on institutions to verify and record an agent’s intended action before it initiates a payment, approves a loan, executes a trade, or files a regulatory report.
For banks already deploying agentic AI across payments, lending, fraud management, and customer servicing, the operational specificity of the recommendation matters. It points directly at a control that is usually missing, as institutions typically provision an agent once with a broad set of permissions and then largely leave it to operate within them.
Related: Levis Refines Premium Strategy in Asia
Asia’s regulators are converging on the same principle, with Hong Kong’s Privacy Commissioner for Personal Data urging organizations to adopt stronger governance and limit agents to the minimum access rights needed for each task.
China’s National Information Security Standardisation Technical Committee released draft guidance covering the deployment, operation, and retirement of AI agents across their lifecycle, taking a broader view of the same underlying risk.
Bank access models were designed around two kinds of users: employees and machine identities such as service accounts and APIs. AI agents fit neither category comfortably, as they act without continuous human involvement and can chain tasks together, retrieve data, and reach systems that were never anticipated when access was first granted.
Related: Japan’s 7.1 Quake May Explain Kumamoto Damage
An agent reviewing loan documentation, for example, may pull customer records, trigger verification checks, and interact with payment or treasury systems in the course of a single task, moving across systems at machine speed in ways no static permission set was designed to anticipate.
As AI agents become more prevalent, banks will need to extend identity governance to these autonomous actors.
Banks that build these controls now, while expectations are still guidance rather than rule, will spend far less time retrofitting once supervisors formalize them. The institutions that benefit most will not be the ones that wait for a single mandatory regulation to emerge, but the ones already extending identity governance to autonomous actors.
