In a recent commentary, Torsten Slok, chief economist at Apollo Global Management, sounded the alarm on a potential new risk to the banking system that stems from the rapid evolution of artificial intelligence. He argues that the emergence of sophisticated AI agents—software programs capable of making autonomous financial decisions—could inadvertently trigger a form of bank run, not through panic or rumor, but via systematic, algorithm‑driven reallocation of funds.
At the heart of Slok’s concern is the way AI agents are being integrated into personal finance management tools, robo‑advisors, and even everyday banking apps. These agents are increasingly given the authority to optimize a user’s cash holdings by searching for the highest possible return on idle money. Traditionally, a household might keep a modest amount of cash in a checking account that earns little or no interest, simply for convenience and to cover day‑to‑day expenses. However, modern AI‑driven platforms can scan a multitude of financial products—high‑yield savings accounts, money‑market funds, short‑term bond ETFs, and even decentralized finance (DeFi) protocols—to identify marginally better yields.
When an AI agent identifies a more attractive rate, it can automatically initiate a transfer, moving the funds from the low‑interest checking account to the higher‑yield vehicle without any manual intervention from the account holder. While this automation offers clear benefits—higher earnings on idle cash, reduced need for active portfolio management, and a more efficient allocation of resources—it also introduces a feedback loop that could have systemic implications.
If a large number of consumers adopt such AI tools, the aggregate effect could be substantial. Imagine millions of checking accounts across the United States each holding an average of $5,000 in low‑interest balances. Even a modest shift of 10 percent of those funds—driven by AI agents seeking a 0.25‑percentage‑point improvement in yield—would represent a massive outflow from traditional deposit accounts.
Because the moves are triggered by algorithms that react to the same market signals, they could occur almost simultaneously, creating a sudden and pronounced reduction in the deposits that banks rely on for funding. Banks traditionally depend on a stable base of low‑cost deposits to fund loans and other assets. A rapid, algorithm‑driven exodus of cash could force banks to scramble for alternative funding, potentially raising borrowing costs or prompting them to sell assets at unfavorable prices.
In extreme scenarios, the loss of deposits could erode confidence in a bank’s liquidity position, prompting regulators to intervene or, in worst‑case scenarios, leading to a cascade of failures reminiscent of classic bank runs, albeit driven by technology rather than human fear. Slok emphasizes that this risk is distinct from the classic panic‑driven runs of the past. Instead of depositors rushing to withdraw cash in a frenzy, the withdrawals would be orderly, automated, and based on rational financial optimization.
Yet the cumulative effect would be the same: a sudden contraction of the funding base that banks count on to meet their obligations. Moreover, because the triggers are embedded in code, the speed and scale of the outflows could outpace traditional monitoring tools, leaving banks with limited time to react.
Regulators and policymakers are beginning to take note. The Federal Reserve, the Office of the Comptroller of the Currency, and other supervisory bodies have started to explore how AI‑driven financial behavior might affect systemic stability. Potential safeguards could include requiring banks to hold higher liquidity buffers, imposing limits on the speed at which large deposits can be moved, or mandating transparency from fintech firms about the algorithms that drive fund transfers. From a consumer perspective, the benefits of AI agents remain compelling.
Automated cash management can boost household wealth over time, especially in a low‑interest environment where every basis point matters. However, education is essential. Users should understand that while their AI assistant is working to maximize returns, it is also participating in a broader financial ecosystem that could be stressed by mass adoption. Industry participants are also exploring ways to mitigate the risk.
Some banks are developing their own AI‑enabled cash‑management services, allowing them to retain the deposit while still offering competitive yields through internal product innovation. Others are collaborating with fintech firms to create “soft‑landing” mechanisms—gradual transfer schedules that spread out movements over days or weeks, reducing the shock to the banking system.
In summary, Torsten Slok’s warning highlights a paradox of modern finance: technology designed to improve individual outcomes may inadvertently create collective vulnerabilities. The key lies in striking a balance between harnessing AI’s efficiency and ensuring that the financial infrastructure can absorb the resulting flows without destabilizing. Ongoing dialogue among banks, fintech innovators, regulators, and consumers will be crucial to develop policies and technical safeguards that preserve the benefits of AI‑driven cash optimization while protecting the stability of the banking sector.