In a recent commentary, Torsten Slok, a senior economist at Apollo Global Management, highlighted a growing concern that advanced artificial‑intelligence agents could inadvertently set off a modern‑day bank run. The crux of his warning revolves around the ability of these AI‑driven tools to automatically reallocate personal cash balances from traditional, low‑interest checking accounts into more lucrative investment vehicles, such as high‑yield savings accounts, money‑market funds, or even short‑term bond ETFs.
While the underlying intention is to maximize returns for consumers, the collective effect of millions of households following the same algorithmic recommendation could drain the reserves of banks that rely on cheap deposits to fund loans and other core activities. Slok’s analysis begins with a description of the evolving financial landscape. Over the past decade, the spread between the interest rates offered on everyday checking accounts and those available on alternative, higher‑yield products has widened dramatically.
Traditional banks, especially large retail institutions, have struggled to compete with fintech firms and online banks that can afford to pay substantially more on deposits because of lower overhead costs. This disparity has already prompted many consumers to move a portion of their cash into better‑paying accounts, but the process has largely been manual, requiring individuals to research options, open new accounts, and transfer funds. Enter AI agents. These are sophisticated software programs that can scan a user’s financial portfolio, assess risk tolerance, and execute transactions without direct human intervention.
Powered by machine learning models and integrated with banking APIs, such agents can continuously monitor interest‑rate changes across the market and instantly shift money to wherever the highest yield is offered. Slok points out that, in theory, this technology promises great benefits: higher earnings for savers, more efficient capital allocation, and reduced friction in personal finance management.
However, the economist warns that the very efficiency of these agents could become a systemic risk. If a large number of households employ similar AI tools—perhaps even the same commercially available product—the agents will all receive the same signals when a higher‑yield option emerges. Their programmed response will be to move funds en masse. In a short window, banks that depend on low‑cost deposits could see a sudden outflow of cash, forcing them to sell assets or seek more expensive funding sources.
The situation mirrors a classic bank run, where fear of insolvency prompts depositors to withdraw their money, thereby creating the very insolvency they feared. Slok emphasizes that the risk is not limited to a single bank. Many institutions share similar funding structures, and the interconnectedness of the financial system means that a liquidity squeeze at one major bank could quickly ripple through the sector. Moreover, the speed at which AI agents can act—often within seconds—means that regulators and bank managers would have far less time to intervene compared with a traditional, slower‑moving run driven by human panic.
To mitigate these dangers, Slok suggests several policy and industry measures. First, greater transparency around the algorithms that drive these agents could help regulators assess systemic exposure. If the decision‑making logic is known, authorities could simulate stress scenarios and identify potential flash‑point events before they happen. Second, banks might need to redesign their deposit contracts to include clauses that limit rapid, large‑scale withdrawals, or to offer tiered interest rates that reward longer‑term balances, thereby reducing the incentive for instant reallocation.
Another recommendation is the development of coordinated industry standards for AI‑driven financial advice. By encouraging diversity in the underlying models—so that not all agents respond identically to market changes—the financial system could avoid a synchronized exodus of funds.
Finally, Slok calls for a dialogue between fintech innovators, traditional banks, and supervisory bodies to craft a regulatory framework that balances innovation with stability. This could involve setting caps on the volume of deposits that can be moved automatically within a given timeframe, or requiring AI providers to implement cooling‑off periods before executing large transfers. In conclusion, while AI agents hold the promise of empowering consumers to earn better returns on their idle cash, the collective behavior they could induce poses a novel threat to banking stability.
Slok’s warning serves as a reminder that technological progress must be accompanied by thoughtful oversight and proactive risk management. By anticipating the ways in which autonomous financial tools might amplify traditional vulnerabilities, policymakers and industry leaders can work together to ensure that the benefits of AI are realized without compromising the resilience of the banking system.