In a recent commentary, Torsten Slok, a senior economist at Apollo Global Management, sounded the alarm about a new, technology‑driven risk to the banking system. He argued that the rapid advancement of artificial‑intelligence‑powered agents—software programs that can autonomously manage personal finances—might inadvertently trigger a wave of deposit withdrawals that resembles a classic bank run, but with a digital twist.
Traditionally, a bank run occurs when a large number of depositors rush to withdraw their money from a financial institution out of fear that the bank may become insolvent. The panic itself can cause the bank’s liquidity to evaporate, turning a solvency concern into an actual crisis.
Slok’s warning does not focus on a sudden loss of confidence in a single bank’s stability; rather, he points to a systematic shift in how consumers allocate cash in response to the ever‑widening gap between the meager interest rates offered on everyday checking accounts and the more attractive yields available on alternative investment vehicles. The core of the concern lies in the capabilities of AI agents that are increasingly being integrated into personal‑finance apps, robo‑advisors, and even voice‑activated assistants.
These agents can continuously monitor a user’s account balances, compare rates across a multitude of platforms, and execute transfers with minimal human intervention. When an AI algorithm detects that a checking account is earning, for example, 0.01 % APY while a short‑term Treasury fund or a high‑yield online savings account offers 4 % or more, the agent can automatically move the excess cash to the higher‑yield option. While such behavior is beneficial for the individual’s personal return, the aggregate effect across millions of households could be profound.
Slok explains that the banking sector’s business model still relies heavily on low‑cost deposits to fund loans and other higher‑margin activities. Even though interest‑rate environments have been low for years, banks have historically been able to retain a stable base of deposits because many consumers keep a portion of their money in checking accounts for convenience, habit, or lack of awareness about better options.
However, AI agents erode those frictions. By eliminating the need for manual research and the inertia that typically keeps cash in low‑yield accounts, these agents could cause a rapid outflow of funds from traditional banks to alternative platforms that promise higher returns. The speed at which AI agents operate is another factor that amplifies the risk. Human depositors might take weeks or months to notice a better rate and to act on it.
An AI‑driven system, on the other hand, can execute a transfer in seconds once the rate differential crosses a predefined threshold. If a large number of agents are programmed with similar thresholds, a modest change in market rates could trigger a cascade of simultaneous withdrawals. This synchronized movement of money could strain a bank’s liquidity buffers, forcing it to sell assets at unfavorable prices or to seek emergency funding from central banks. Moreover, the phenomenon could be self‑reinforcing.
As banks experience deposit outflows, they may be compelled to raise interest rates on checking accounts to retain customers, which in turn could narrow the yield gap and reduce the incentive for AI agents to move money. Conversely, if banks are unable or unwilling to adjust rates quickly, the outflows could continue unabated, potentially leading to a feedback loop where declining deposits force banks to cut back on lending, which then slows economic activity and further undermines confidence.
Slok also highlights that the risk is not confined to any single type of financial institution. While large, well‑capitalized banks might have the resources to weather short‑term liquidity shocks, smaller regional banks and community banks—often more dependent on a stable deposit base—could be disproportionately affected. A sudden, technology‑driven shift of even a modest portion of their deposits could push these institutions into a precarious position, especially if they lack diversified funding sources. Regulators are beginning to take notice of this emerging dynamic.
The Federal Reserve and other supervisory bodies have traditionally focused on liquidity ratios, stress‑testing, and capital adequacy. However, the rise of AI agents introduces a new variable that is not easily captured by conventional metrics.
Slok suggests that regulators may need to incorporate scenario analyses that model the rapid, automated reallocation of deposits in response to rate differentials, as well as monitor the adoption rates of AI‑driven financial tools. From a consumer‑protection standpoint, there are also questions about transparency and consent. Many users may not be fully aware that an AI assistant is making autonomous decisions about where their cash is parked. While the financial benefit is clear, the lack of explicit user involvement could raise concerns about the adequacy of disclosures and the potential for unintended consequences, such as moving funds into products that carry higher risk or lower liquidity.
In light of these considerations, Slok recommends a multi‑pronged approach. First, banks should enhance the attractiveness of their deposit products, not only through higher rates but also by offering value‑added services that AI agents might struggle to replicate, such as personalized financial planning or integrated payment solutions.
Second, financial institutions could develop their own AI capabilities to retain control over the flow of deposits, perhaps by offering customers AI‑assisted tools that keep money within the bank while still optimizing returns. Third, policymakers should consider updating regulatory frameworks to capture the speed and scale at which AI agents can move funds, possibly introducing reporting requirements for large‑scale automated transfers.
Ultimately, the issue underscores a broader theme: technology is reshaping the fundamentals of banking in ways that extend beyond cybersecurity and digital onboarding. The automation of everyday financial decisions, while empowering for individual savers, introduces systemic dynamics that could challenge the traditional stability of the banking sector. As Slok warns, without proactive measures from both banks and regulators, the collective action of AI agents could unintentionally create a modern, algorithm‑driven version of a bank run, draining cheap deposits that have long served as the backbone of the industry.
The conversation around AI‑driven deposit migration is still in its early stages, but the potential implications are significant. Stakeholders across the financial ecosystem—banks, fintech firms, regulators, and consumers—must engage in a dialogue to balance the benefits of automation with the need to preserve liquidity stability. By anticipating these shifts and adapting strategies accordingly, the industry can harness the power of AI without compromising the foundational trust that underpins the banking system.