913b2cac-2

ConfirmedRegulation & Policy614 words3 min decode

Singapore MAS plans agent identity checks and AI trading caps

Singapore’s MAS is pursuing “Know Your Agent” checks and limits on autonomous AI trading while testing tokenized deposits, stablecoins and AI payments in finance.

Singapore’s MAS Warns AI Agents May Heighten Financial Stability Risks, Plans Transaction-Authority Checks
WitnessSingapore’s MAS Warns AI Agents May Heighten Financial Stability Risks, Plans Transaction-Authority ChecksDenisStreltsov / pixabay

Outputs

  1. On November 19, Bloomberg reported that MAS Managing Director Chia Der Jiun urged financial institutions to cap AI trading instructions.

  2. MAS is pursuing a “Know Your Agent” protocol covering agent identity, user authority and permitted scope.

  3. In November, MAS issued S$100 million of tokenized short-term bonds settled in a Singapore dollar CBDC.

  4. MAS plans AI-payment tests in which agents monitor liquidity and pay with tokenized deposits or stablecoins.

  5. The announcement set no numerical AI trading cap, rulemaking deadline or implementation timetable.

Daniel Okafor

Singapore’s Monetary Authority of Singapore (MAS) is pursuing a “Know Your Agent” protocol and limits on autonomous AI trading, warning that rapid agent-driven fund transfers could raise liquidity risk in the financial system.

In a Bloomberg report published November 19, MAS Managing Director Chia Der Jiun said AI agents can move money and allocate assets faster and with smaller incentives.

Speaking in a keynote address at the Singapore FinTech Festival, he called on financial institutions to cap AI trading instructions and urged regulators to monitor the technology’s implications for financial stability.

Why is MAS raising the warning?

MAS said it would strengthen risk management in financial transactions through those two safeguards. The common framework would give banks and payment providers one way to verify an AI agent’s identity, the authority granted by a user and the limits of its mandate before the agent acts.

Chia tied the warning to the speed of agent activity. His call for caps would constrain trading instructions from autonomous systems, although the announcement did not disclose a numerical ceiling or an implementation date.

What would “Know Your Agent” verify?\n

At the center of MAS’s proposal is a common method for financial firms to establish three things:

  • The identity of the AI requesting a transaction.
  • The authority the user has granted that AI.
  • How far the AI may act on the user’s behalf.

For banks and payment providers, the proposal would shift the compliance focus from possession of account credentials to the validity and limits of a user’s delegation. An authenticated agent could not act beyond its mandate simply because it could reach the customer’s account.

Chia also urged caution about individuals giving login credentials for financial accounts to AI agents, saying the safety of that approach had not been sufficiently assured. That separates authentication of the agent from the risk created by exposing account access.

How will MAS test agent payments?

MAS plans to continue experiments involving AI-enabled financial transactions across tokenized deposits, stablecoins, a Singapore dollar central bank digital currency and tokenized short-term bonds. In the payment tests, agents will monitor liquidity conditions and make payments using tokenized deposits or stablecoins.

The announcement did not identify a stablecoin, token standard or blockchain network for those tests.

The program’s broader aim is to expand AI use in finance while putting identity checks, authority limits and accountability frameworks in place. Chia said: “The future is likely to be a mix of a human economy and a new agent economy.” He added: “Autonomy must have boundaries, and actions and outcomes must carry accountability.”

What infrastructure supports the model?

MAS is also building out digital financial infrastructure. In November, it issued S$100 million of tokenized short-term bonds settled in a Singapore dollar central bank digital currency. The issuance pairs tokenized short-term debt with central-bank digital currency settlement in MAS’s infrastructure program.

MAS plans to work with the industry to use major collateral assets, including foreign government bonds and money market funds, on digital platforms. The initiative would connect digital infrastructure with assets MAS classifies as major collateral, while the AI experiments test how agents monitor liquidity conditions and complete tokenized payments.

When will safeguards become binding?

The announcement did not provide a rulemaking deadline, numerical trading cap or implementation timetable. For institutions, the proposed model would require compliance teams to connect customer mandates with payment controls rather than treat account access as sufficient authorization.

The next market-structure checkpoint will be whether MAS turns the proposed protocol and unspecified caps into common, enforceable rules before banks and payment providers scale agent-led transactions.

via en.bloomingbit.io (Source)

More from Daniel Okafor

Daniel Okafor

Show full bio

Correspondent covering people, launches and day-to-day practice in crypto and web3 industry at Mempool Brief.

1,584 articles