In a recent commentary, Torsten Slok, the chief economist at Apollo Global Management, raised a cautionary flag about the emerging role of artificial‑intelligence‑driven agents in the financial system. He argued that as these autonomous programs become more sophisticated, they could inadvertently trigger a rapid outflow of funds from traditional, low‑interest deposit accounts, effectively creating a modern‑day bank run.
While the concept may sound like something out of a science‑fiction thriller, the underlying mechanics are rooted in very real economic incentives and technological capabilities that are already taking shape. At the heart of Slok’s warning is the simple observation that most consumers keep a substantial portion of their liquid cash in checking accounts that pay little or no interest. Historically, the convenience of easy access to funds has outweighed the modest returns that these accounts provide. However, the financial landscape has begun to shift dramatically in recent years.
A growing array of fintech platforms, high‑yield savings products, and digital‑only banks now offer substantially higher rates on cash that remains readily accessible. For a tech‑savvy individual, the decision to move money from a traditional checking account to a higher‑yield alternative is often a straightforward calculation: higher return for the same level of liquidity. Enter AI agents—software programs that can monitor financial markets, compare product offerings, and execute transactions automatically on behalf of users.
These agents are already being deployed in portfolio management, expense tracking, and even in negotiating better terms on loans. Slok points out that once an AI agent is programmed to optimise cash holdings for maximum yield, it will naturally begin to shift money from low‑interest checking balances into the most attractive alternatives it can locate. This process, when scaled across millions of households, could generate a sudden and massive reallocation of deposits. The risk, according to Slok, lies in the speed and coordination of these automated moves.
Human depositors typically act based on personal schedules, emotional responses, or limited information, which spreads out withdrawals over time. An AI‑driven system, however, can scan thousands of accounts in seconds, identify the best yields, and execute transfers almost instantaneously. If a large number of agents receive the same signal—say, a new high‑yield product is announced or an existing offering raises its rate—they could collectively initiate a wave of transfers within minutes. The resulting surge in outflows could strain the liquidity buffers of banks that rely heavily on low‑interest deposits to fund loans and other assets.
Such a scenario mirrors the classic definition of a bank run, where depositors rush to withdraw funds out of fear that a bank may become insolvent. In this AI‑augmented version, the fear is not necessarily about solvency but about opportunity cost: the perceived loss of higher earnings if cash remains idle in a low‑yield account. Nonetheless, the practical effect is the same—banks experience a rapid depletion of cash reserves, forcing them to sell assets, borrow on the interbank market, or even seek emergency funding from central banks. Slok also highlights that the problem is not limited to a single institution.
Because AI agents can operate across multiple banks and fintech platforms, the pressure could be distributed but simultaneous, creating a systemic liquidity shock. Moreover, the feedback loop could intensify the situation: as banks respond by raising rates on their own checking products to retain deposits, AI agents would immediately reassess the new landscape, potentially shifting funds back and forth in a volatile cycle.
Regulators and policymakers, therefore, need to anticipate these dynamics before they materialise on a large scale. Slok suggests several possible mitigations.
First, banks could implement throttling mechanisms that limit the frequency or volume of automated transfers, giving them more time to manage liquidity. Second, transparency around the algorithms used by AI agents could be mandated, allowing regulators to monitor aggregate behaviour and intervene if a coordinated sweep appears imminent.
Third, encouraging the development of “deposit insurance” products that cover not just traditional bank deposits but also funds held in high‑yield fintech accounts could reduce the perceived risk and dampen the speed of withdrawals. From a consumer perspective, the rise of AI agents offers undeniable benefits—greater efficiency, better returns, and reduced manual effort. However, Slok warns that the collective impact of many individuals pursuing the same optimisation goal can create unintended macro‑economic consequences. The key takeaway is that while technology can enhance personal finance, it must be integrated with safeguards that preserve the stability of the broader financial system.
In summary, Torsten Slok’s alert serves as a reminder that the convergence of AI and finance is not merely a matter of convenience; it carries systemic implications that demand proactive oversight. As AI agents become more capable of autonomously reallocating cash, the traditional safety net provided by low‑interest checking accounts may erode, potentially precipitating a rapid, technology‑driven shift of deposits.
Stakeholders—including banks, fintech firms, regulators, and consumers—must work together to design policies and technical controls that balance the pursuit of higher yields with the need to maintain liquidity and confidence in the banking sector. By addressing these challenges now, the financial industry can harness the power of AI without exposing itself to the risk of a new kind of bank run.