MAS Warns AI Agents Could Move Money Faster, Raise Risks

Singapore's Monetary Authority has issued a formal warning that artificial intelligence agents could accelerate deposit movements and asset reallocation, introducing unprecedented liquidity challenges for financial institutions. Speaking at the Singapore FinTech Festival on November 19, 2026, MAS Managing Director Chia Der Jiun announced plans for a new verification framework as banks worldwide grapple with autonomous AI systems gaining control over customer funds.
Speed and Scale of AI Agent Transactions
Chia highlighted the core risk: agents can execute financial transactions faster and for smaller economic incentives than human users. This capability fundamentally changes how quickly capital can flow through the banking system, potentially destabilizing traditional liquidity management strategies that assume human decision-making timelines.
The concern is not theoretical. Charles Schwab Corp CEO Rick Wurster revealed in October 2026 that the brokerage had already blocked external AI agents attempting to initiate money transfers for customers. Financial institutions are now racing to establish guardrails before widespread AI agent adoption creates systemic vulnerabilities.
New Know Your Agent Protocol
MAS will develop a common verification standard that Chia described as a "know your agent" protocol. This framework will require banks and payment providers to authenticate three critical elements before allowing AI agents to execute transactions: the agent's identity, its authority level, and its specific mandate from the account holder.
The authority explicitly cautioned against giving personal AI agents direct login credentials to financial accounts, citing insufficient safety assurances. Instead, MAS plans to test agents that monitor liquidity positions and execute payments using tokenized deposits or stablecoins, creating a controlled environment for autonomous financial operations.
Bounded Autonomy and Accountability
Chia emphasized that the future economy will blend human and AI agent activity, but autonomy must have clear limits. Financial institutions may need to implement transaction caps or velocity limits on AI agent-instructed operations to manage risk exposure effectively.
Regulators face the challenge of determining liability frameworks when agents make suboptimal or erroneous financial decisions. MAS insists that people and businesses must retain meaningful control over agent actions, with outcomes tied to accountability structures that mirror existing consumer protection standards.
Tokenization Infrastructure Expansion
Officially launched on November 19, 2026, MAS issued 100 million Singapore dollars ($78 million USD) in tokenized MAS Bills settled through Singapore-dollar central bank digital currency. This milestone advances the authority's broader strategy to digitize capital market infrastructure.
The central bank plans to collaborate with industry participants to tokenize additional collateral types, including foreign government securities and money-market funds. Singapore joins Switzerland and the Czech Republic in a small cohort of nations actively deploying distributed ledger technology for monetary policy operations and short-term securities issuance.
What This Means
The MAS warning signals a critical inflection point for AI agents in finance. As technology companies and banks accelerate AI agent deployment for shopping, payments, and wealth management, regulators are establishing the first formal frameworks to govern autonomous financial systems. The know your agent protocol could become a global template, similar to know your customer standards, shaping how AI agents interact with banking infrastructure for years. Financial institutions that fail to implement proper verification and limits may face regulatory action or customer liability issues as agent-driven transactions become mainstream in 2027 and beyond.
on Emergent today
.avif)




