In a recent commentary, Torsten Slok, chief economist at Apollo Global Management, warned that the rise of autonomous artificial‑intelligence agents could fundamentally alter the way individuals manage their cash, potentially creating a new kind of bank‑run scenario. While traditional bank runs have historically been driven by panic, rumors, or sudden loss of confidence in a financial institution, Slok suggests that a more subtle, technology‑driven mechanism could emerge: AI‑powered software that continuously monitors the interest rates offered on checking accounts, savings accounts, money‑market funds, and a growing array of alternative cash‑management products, and then automatically reallocates funds to the highest‑yielding option available. The core of Slok’s argument rests on the fact that many households keep a substantial portion of their liquid wealth in low‑interest checking accounts, often earning near‑zero returns. In a low‑rate environment, the opportunity cost of holding cash in such accounts can be significant, especially for savers who are looking for any edge to improve their net worth.
As AI agents become more sophisticated and more widely adopted, they will be able to scan the market in real time, compare rates across dozens of banks and fintech platforms, and execute transfers without human intervention. This capability, while beneficial for individual investors seeking higher yields, could collectively generate a rapid outflow of deposits from traditional banks. To understand the potential magnitude of this risk, consider a simplified model.
Suppose a mid‑size bank holds $50 billion in checking deposits that earn an average interest rate of 0.05 %. If AI agents detect a competing product offering 0.30 % on a comparable cash‑equivalent vehicle, they may trigger a transfer of even a modest 5 % of each customer’s balance. That would amount to $2.5 billion moving out of the bank in a very short period—potentially days or even hours—depending on the speed of the automation.
Multiply this scenario across hundreds of banks, and the aggregate effect could be a multi‑trillion‑dollar shift in liquidity. Slok emphasizes that the danger is not merely theoretical. The financial system already relies on a stable base of deposits to fund loans, mortgages, and other credit products.
When deposits evaporate quickly, banks may be forced to sell assets at fire‑sale prices, raise short‑term borrowing costs, or, in the worst case, become insolvent. The Federal Reserve’s liquidity facilities and the traditional deposit insurance scheme are designed to address sudden runs, but they may be less effective when the run is driven by algorithmic, pre‑programmed behavior rather than a panic‑induced rush of depositors. Moreover, the speed and scale of AI‑driven transfers could outpace the existing regulatory safeguards.
Current stress‑testing frameworks assume that deposit outflows will follow historical patterns, typically modeled as a gradual decline over weeks or months. An AI‑orchestrated exodus, however, could compress that timeline dramatically, leaving banks with insufficient time to raise capital or access emergency funding. This mismatch could exacerbate systemic risk, especially for smaller regional banks that lack the diversified funding sources of larger institutions.
In addition to the direct impact on banks, Slok points out several secondary consequences. First, the competitive pressure could accelerate the migration of cash to fintech platforms that offer higher yields, such as online money‑market funds, short‑term bond ETFs, or even blockchain‑based stablecoins that promise near‑instant settlement and attractive returns. Second, the rapid reallocation of funds could distort price signals in money markets, as the demand for short‑term, high‑yield instruments spikes unexpectedly.
Finally, there is a potential feedback loop: as more AI agents act on the same data, their collective behavior may amplify market movements, creating volatility that could spill over into broader asset classes. To mitigate these risks, Slok recommends that regulators and banks take a proactive stance.
One suggestion is to develop “AI‑aware” liquidity monitoring tools that can detect patterns of automated withdrawals and flag them before they become systemic. Another approach is to encourage banks to offer more competitive rates on basic deposit products, thereby reducing the incentive for AI agents to move funds elsewhere. Additionally, policymakers could consider updating deposit insurance limits or creating temporary liquidity backstops specifically designed for algorithm‑driven outflows. From a technological perspective, banks themselves could adopt AI to counterbalance the threat.
By deploying their own intelligent agents, banks could personalize offers, dynamically adjust rates, or provide bundled services that make staying with the institution more attractive. For example, an AI‑driven loyalty program could reward customers for maintaining a certain balance, or offer instant, fee‑free transfers to other accounts within the same banking ecosystem, thereby reducing the appeal of external platforms.
In conclusion, Torsten Slok’s warning underscores a subtle but profound shift in the dynamics of deposit stability. As AI agents become ubiquitous in personal finance, the traditional assumption that depositors will act slowly and rationally may no longer hold.
Instead, the market could see a cascade of automated, high‑speed reallocations that challenge the resilience of banks and the broader financial system. Stakeholders—regulators, banks, fintech firms, and consumers—must recognize this emerging risk and collaborate on solutions that preserve liquidity, maintain confidence, and harness the benefits of AI without exposing the system to new, technology‑driven vulnerabilities.