In a recent commentary, Torsten Slok, chief economist at Apollo Global Management, sounded the alarm on a new kind of systemic risk that could emerge from the rapid adoption of artificial‑intelligence‑powered financial agents. While AI has already transformed many aspects of banking—from fraud detection to personalized product recommendations—Slok points out that the next wave of autonomous software could inadvertently trigger a modern‑day bank run. The mechanism, he explains, is surprisingly straightforward: sophisticated AI agents, programmed to optimize the return on every dollar they manage, will automatically scan a household’s portfolio, identify low‑interest checking balances, and reallocate those funds to higher‑yield alternatives such as money‑market funds, short‑term bond ETFs, or even decentralized finance platforms.
If millions of consumers employ similar agents, the cumulative effect could be a sudden, large‑scale outflow of cash from traditional deposit accounts. The concern is not merely theoretical. Over the past decade, the United States has witnessed a steady decline in the average interest rate paid on ordinary checking accounts, often hovering near zero percent. At the same time, the proliferation of fintech apps and robo‑advisors has made it easier than ever for individuals to move money with a few taps on a smartphone.
When these tools are augmented with machine‑learning algorithms that continuously monitor market yields, the incentive to chase even marginally better returns becomes a built‑in driver of behavior. Slok argues that once the technology reaches a level of sophistication where agents can act without human oversight—executing trades, reallocating assets, and even negotiating terms with counterparties—the speed and scale of fund migrations could outpace the ability of banks to manage liquidity. Historically, bank runs have been triggered by panic, rumor, or a loss of confidence in a particular institution.
The classic image is a line of depositors rushing to withdraw cash from a teller window, forcing the bank to liquidate assets at fire‑sale prices. In the digital age, the physical line is replaced by algorithmic orders that can be submitted in milliseconds. An AI agent that detects a decline in a bank’s net interest margin or a downgrade in its credit rating could instantly instruct its client’s account to move funds elsewhere. If a critical mass of agents makes the same decision at roughly the same time, the resulting outflow could be mathematically equivalent to a traditional run, albeit executed on a digital ledger.
Slok emphasizes that the risk is amplified by two interrelated trends. First, the growing concentration of wealth in a few large technology platforms gives those platforms the power to embed AI agents directly into their ecosystems. For example, a major smartphone operating system could integrate a financial assistant that automatically optimizes cash balances across all linked accounts. Second, the regulatory environment has not yet caught up with the speed at which these agents operate.
Current liquidity requirements for banks are based on assumptions about human‑driven withdrawals, which are typically spread out over days or weeks. Real‑time, algorithmic withdrawals could render those safeguards insufficient. To mitigate this emerging threat, Slok suggests a multi‑pronged approach. Regulators could require banks to disclose the proportion of deposits that are managed by autonomous agents, thereby providing a clearer picture of potential liquidity stress points.
Additionally, stress‑testing frameworks might be updated to simulate scenarios where AI‑driven outflows occur within minutes rather than months. On the technology side, developers of financial AI should incorporate safeguards that limit the frequency or volume of automatic transfers, perhaps by imposing a minimum holding period or requiring a manual confirmation for large moves. From a consumer perspective, awareness is key. While the promise of higher yields is attractive, individuals should understand that the very tools designed to maximize returns could also contribute to systemic instability.
Diversifying across multiple institutions, maintaining a modest buffer of readily accessible cash, and periodically reviewing the settings of any automated financial assistant can help mitigate personal exposure. In conclusion, the convergence of low‑interest deposit environments, highly efficient AI agents, and the digitalization of banking creates a perfect storm that could lead to rapid, large‑scale withdrawals from traditional banks. Torsten Slok’s warning serves as a reminder that innovation, while beneficial, must be paired with thoughtful oversight and prudent risk management. By recognizing the potential for AI‑induced liquidity shocks early, policymakers, financial institutions, and consumers alike can take steps to ensure that the benefits of artificial intelligence do not come at the expense of financial stability.