In a recent commentary, Torsten Slok, the chief economist at Apollo Global Management, sounded the alarm about a new, technology‑driven risk to the stability of the banking system. He explained that sophisticated artificial‑intelligence agents, designed to optimize personal finance, could unintentionally act as a catalyst for a modern‑day bank run.
By continuously scanning the market for the best possible return on cash, these AI‑powered tools might automatically transfer money out of traditional, low‑interest checking accounts and into higher‑yield products such as money‑market funds, short‑term bond ETFs, or even emerging crypto‑based savings platforms. Slok’s warning hinges on the fact that many households keep a substantial portion of their liquid assets in checking accounts that offer virtually no interest.
While these accounts provide convenience and safety, they are increasingly unattractive compared to alternative vehicles that promise better yields with comparable liquidity. As AI assistants become more adept at personal finance management, they are programmed to seek out the most efficient allocation of a user’s cash.
In practice, this means that when a new, slightly higher‑yielding product appears, the AI could instantly recommend, or even execute, a transfer of funds without waiting for the account holder to manually intervene. The potential problem emerges when millions of such agents operate in concert.
If a large number of AI tools simultaneously identify a marginally better rate and move funds en masse, the aggregate effect could be a rapid outflow of deposits from traditional banks. This scenario mirrors the classic definition of a bank run, where depositors rush to withdraw their money out of fear that the institution may become insolvent.
However, unlike historical runs that were driven by panic and rumor, this new phenomenon would be algorithm‑driven, systematic, and potentially far more efficient. Slok points out that the speed and scale of these AI‑initiated transfers could outpace the ability of banks to respond. Traditional banks rely on a relatively stable deposit base to fund loans and other assets. A sudden, large‑scale shift of cash to higher‑yield alternatives could force banks to liquidate assets quickly, possibly at a loss, to meet withdrawal demands.
This could compress profit margins, raise funding costs, and ultimately erode confidence in the banking sector. Moreover, the risk is amplified by the fact that many AI agents are integrated with open‑banking APIs, which grant them direct access to a user’s accounts.
When permission is granted, the AI can move funds with minimal friction, often requiring only a single click—or no click at all—if the user has opted into automated optimization. This level of integration, while convenient, creates a conduit through which large sums can be redirected in a matter of seconds. To illustrate the potential impact, consider a hypothetical scenario: a popular budgeting app incorporates an AI module that continuously monitors interest rates across a range of financial products.
The AI detects that a newly launched online savings account offers a 0.75% annual percentage yield (APY), compared to the 0.01% APY on a user’s existing checking account. The AI automatically initiates a transfer of the user’s $10,000 balance to the higher‑yield account.
If one million users of the app each hold an average of $5,000 in checking accounts, the cumulative outflow could amount to $5 billion in a very short period. Such a massive shift would be felt across the banking system, potentially triggering liquidity strains for institutions that depend heavily on retail deposits.
Slok also emphasizes that the problem is not limited to a single AI platform. The financial technology ecosystem is increasingly crowded, with numerous fintech firms offering AI‑driven financial advice, robo‑advisors, and automated savings tools. Each of these platforms could independently decide to move funds based on similar signals, creating a feedback loop that magnifies the overall effect. Regulators are beginning to take notice.
In the United States, the Federal Reserve and the Office of the Comptroller of the Currency have started to explore how AI‑enabled financial services might affect systemic risk. They are considering whether existing safeguards—such as liquidity coverage ratios and stress‑testing frameworks—need to be updated to account for the speed and coordination possible through algorithmic behavior. Potential mitigations could include imposing limits on the frequency or volume of automated transfers, requiring explicit user confirmation for each move, or developing industry‑wide standards for AI‑driven fund allocation.
Banks themselves might respond by enhancing the attractiveness of their own deposit products, offering higher rates or additional perks to retain customers. Some institutions are already experimenting with “AI‑friendly” accounts that integrate directly with popular budgeting tools while maintaining a level of control over how and when funds can be moved. In summary, Torsten Slok’s cautionary message underscores a subtle yet profound shift in how technology can influence financial stability. While AI agents promise greater efficiency and higher returns for individual savers, their collective behavior could unintentionally create a new form of systemic vulnerability.
Stakeholders—including fintech developers, banks, regulators, and consumers—must work together to design safeguards that preserve the benefits of AI‑driven finance without compromising the resilience of the broader banking system.