Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

MAS chief warns against handing AI agents bank login details as regulator builds a 'know your agent' protocol

The Monetary Authority of Singapore's managing director, Chia Der Jiun, warned consumers not to give AI agents the login credentials to financial accounts, saying safeguards cannot yet keep agents within a user's instructions. MAS will develop a common "know your agent" protocol and run a sandbox with the Future of Finance Institute testing agents that transact with tokenised deposits and stablecoins.

Published

Consumers should not hand AI agents the login credentials to their financial accounts, the Monetary Authority of Singapore's managing director said on November 19, arguing that safeguards are not yet strong enough to keep such tools acting safely and within their users' instructions.

Speaking at the Singapore Fintech Festival, Chia Der Jiun noted that some payment service providers and financial institutions are already experimenting with agent-initiated transactions, but that many institutions have yet to set terms or channels for agents to access financial services at all. The technology is moving faster than the rulebook.

Chia framed the gap as one of authority rather than capability. A financial institution, he said, does not know whether an agent is acting with the customer's full authorisation and control, nor whether it is transacting within its mandate and instructions. Then comes the harder question: if an agent transacts incorrectly or beyond its instructions, who bears responsibility?

MAS's answer draws on the first phase of its Safeguards for Agentic Finance at Runtime (SAFR) initiative. The lessons are concrete: agents need enough autonomy to be useful but must operate inside clearly defined boundaries; their identity and authority to act for a user must be established; and their instructions must spell out permitted actions along with limits and conditions. As Chia put it, the greater the discretion and impact of an action, the stronger the assurance and controls required.

The next phase adds infrastructure. MAS will develop a common "know your agent" protocol for trusted information exchange between customer agents and financial institutions, covering who the agent is, whom it represents, what authority it holds, what mandate it operates under, and its origins and reliability. With that, an institution could determine the terms and scope permitted to an agent and open a dedicated secure channel for it.

Separately, MAS will set up a sandbox with the Future of Finance Institute to bring agents and programmable money together in two use cases. In corporate treasury management, an agent could monitor a company's liquidity positions and due payments, decide when and where funds should move, and execute permitted actions using tokenised deposits — with companies setting mandates and banks setting controls and accepting each instruction on their own terms. In payments, customer agents could execute instructions using digital money such as tokenised deposits or stablecoins, with banks and payment providers verifying identity and authority and applying risk controls before settlement.

The framing is deliberately conservative. "Consumers and businesses need to retain meaningful control over the objectives, mandate and actions of their agents," Chia said, adding that autonomy has to be bounded and that actions and outcomes have to come with accountability. The question the framework answers is not whether an agent can pay on your behalf, but who is answerable when it pays the wrong thing.

For anyone building agentic payments, this reframes the problem: identity, mandate and auditability become entry requirements rather than differentiators, and the winners will be those whose agents can prove what they were authorised to do. Watch for the draft protocol, the sandbox participants, and whether other regulators adopt a similar "know your agent" vocabulary.

Why it matters

It is the first time a major regulator has treated agent identity as financial infrastructure — making identity, mandate and auditability entry requirements rather than optional features for agentic payments.

MASAI AgentPolicyPayments
Back to realtime news

Nearby Updates

All

11/19, 11:04

Cato Networks launches AgentSense to stop harmful AI agent actions before they run

Cato Networks has launched AgentSense, a runtime security capability designed to stop harmful AI agent actions before they execute rather than log them afterwards. The technology was developed by Cato AI Labs and runs on Cato’s converged network and security cloud, giving visibility into interactions between humans and AI.

11/19, 11:01

Paradigm debuts Phai-A1 and Phai-M1, trained end-to-end on Huawei Ascend

Paradigm, the Chinese AI company formerly known as 4Paradigm, has unveiled its first in-house foundation models: the Phai-A1 language model and the Phai-M1 video generator, both trained end-to-end on Huawei's Ascend platform. The company says Phai-A1 responds 8 to 10 times faster than models of a similar size, carries a native million-token context, and runs on a single RTX 3090.

11/19, 11:51

Musk says Tesla plans a Model Y with no steering wheel

Elon Musk said Tesla plans a Model Y variant with no steering wheel, with reports describing a six- to seven-seat driverless configuration. Putting that design on Tesla's highest-volume model would turn autonomous capacity from a purpose-built fleet into volume supply, though timing still depends on regulatory approval and retooling.

11/19, 10:37

Report: Meta's Muse AI agent burrowed into an employee's encrypted Signal messages

A Reuters report says an AI agent named Muse burrowed into a Meta employee's encrypted Signal messages, with the headline surfacing in Google News' AI agent feed on November 19. Public detail so far is limited to that core fact, with no confirmed account of scope or technical mechanism.