In a recent commentary, Torsten Slok, a senior economist at Apollo Global Management, highlighted a potential new risk to the banking sector that stems from the rapid advancement of artificial intelligence. According to Slok, the emergence of sophisticated AI agents—software programs capable of making autonomous financial decisions—could inadvertently trigger a large‑scale withdrawal of funds from traditional, low‑interest deposit accounts. This scenario, he argues, bears a striking resemblance to a classic bank run, albeit one driven by technology rather than panic or rumor.
The core of Slok’s warning centers on the way AI agents are being integrated into personal finance management tools. Many fintech platforms now offer AI‑powered assistants that monitor a user’s cash balances, compare interest rates across a range of products, and automatically reallocate money to maximize returns. While this automation promises greater efficiency and higher yields for individual savers, it also creates a feedback loop that could amplify the movement of funds away from conventional checking accounts, which typically offer minimal interest.
In the traditional banking model, a substantial portion of a bank’s funding comes from low‑cost deposits—checking accounts, savings accounts, and other short‑term instruments that provide a stable, inexpensive source of capital. These deposits enable banks to extend loans, support liquidity, and fund other financial activities at a relatively low cost. However, when AI agents continuously scan the market for better rates and shift money to higher‑yield alternatives—such as money‑market funds, short‑term Treasury securities, or digital‑only banks offering attractive promotional rates—the supply of cheap deposits can erode rapidly.
Slok points out that this erosion could happen almost instantaneously. Unlike human depositors who might take weeks or months to move their money after hearing about a better rate, AI agents can execute transactions in seconds. If a new, higher‑yield product is launched, an AI‑driven portfolio manager could instantly rebalance a client’s holdings, pulling cash out of a traditional checking account and placing it into the new vehicle.
Multiply this behavior across millions of households, and the cumulative effect could be a sudden, large‑scale outflow of deposits. The implications for banks are significant. First, a rapid decline in low‑cost deposits forces banks to seek more expensive funding sources, such as wholesale borrowing or issuing higher‑interest certificates of deposit.
This shift raises banks’ overall funding costs, compresses net interest margins, and can diminish profitability. Second, the sudden liquidity drain may strain a bank’s ability to meet short‑term obligations, potentially prompting regulatory scrutiny or even emergency liquidity measures. Moreover, Slok warns that the phenomenon could create a self‑fulfilling prophecy. As banks experience higher funding costs, they may raise rates on their own deposit products to retain customers, which could further incentivize AI agents to seek even better alternatives.
The cycle could accelerate, leading to heightened volatility in the banking sector’s funding structure. To mitigate these risks, Slok suggests that regulators and financial institutions need to adapt their oversight frameworks. He recommends that banks monitor the usage patterns of AI‑driven financial tools among their customers and develop stress‑testing scenarios that incorporate rapid, technology‑induced deposit withdrawals.
Additionally, banks might consider offering more competitive rates on certain low‑interest accounts or bundling value‑added services that are less easily replicated by AI agents. From a broader perspective, Slok’s analysis underscores a fundamental shift in how technology can influence macro‑financial stability. While AI has the potential to enhance financial inclusion, improve asset allocation, and deliver personalized advice, it also introduces novel systemic risks that were not present in the era of purely human decision‑making.
The speed and scale at which AI can act demand a reevaluation of traditional risk‑management tools and regulatory safeguards. In conclusion, Torsten Slok’s cautionary note serves as a reminder that the benefits of AI in finance must be balanced against the potential for unintended consequences. As AI agents become more prevalent in everyday money‑management tasks, banks, policymakers, and fintech developers will need to collaborate closely to ensure that the financial system remains resilient, even as the underlying mechanisms of deposit behavior evolve dramatically.