In a recent commentary, Torsten Slok, the chief economist at Apollo Global Management, raised a warning that the rapid adoption of artificial‑intelligence‑driven financial agents could fundamentally reshape the way ordinary consumers manage their cash, potentially creating a new form of bank run. While the term "bank run" traditionally evokes images of panicked crowds lining up at teller windows, Slok suggests that the same destabilizing effect could occur in the digital age, driven not by physical queues but by algorithms that automatically reallocate money from low‑interest checking accounts to more lucrative investment vehicles. The core of Slok’s argument rests on the observation that many households keep a substantial portion of their liquid assets in checking accounts that pay little or no interest.
Historically, these accounts have served as a convenient place to store cash for day‑to‑day expenses, with the implicit understanding that the funds are safe and readily accessible. However, the rise of fintech platforms, high‑yield savings accounts, and low‑cost money‑market funds has introduced a plethora of alternatives that can offer significantly higher returns on idle cash. In response, a new generation of AI‑powered financial assistants—sometimes referred to as "robo‑advisors" or "AI agents"—has emerged, promising to optimize personal finances by continuously seeking the best possible yields. According to Slok, the danger lies in the speed and scale at which these AI agents can act.
Unlike a human who might manually move money after conducting research and weighing risk, an algorithm can execute transfers in seconds, scanning multiple platforms, comparing rates, and reallocating funds with minimal human oversight. If a sizable segment of the population adopts such agents, a coordinated shift of deposits could occur almost simultaneously whenever a more attractive rate surfaces. The result would be a rapid outflow of funds from traditional checking accounts, which banks rely on as a cheap source of capital. Banks typically fund a large portion of their lending and investment activities with these low‑cost deposits.
When deposits dwindle, banks may be forced to raise the interest rates they pay to retain customers, increase borrowing costs, or even sell assets to meet liquidity requirements. In extreme cases, a sudden, large‑scale withdrawal of deposits—whether orchestrated by humans or triggered by autonomous agents—could strain a bank’s balance sheet, prompting regulatory intervention or, in the worst‑case scenario, a failure. Slok emphasizes that this risk is not merely hypothetical.
He points to recent trends where fintech firms have attracted billions of dollars in deposits by offering rates that far exceed those of traditional banks. For example, online-only banks and neobanks have been able to advertise yields of 4% to 5% on savings products, while many major brick‑and‑mortar institutions still offer rates below 1% on checking balances.
The disparity creates a strong incentive for AI agents to continuously chase the highest yield, especially when the agents are programmed to prioritize net returns after fees and taxes. Moreover, the regulatory environment may lag behind the speed of technological innovation. Current banking regulations, such as reserve requirements and liquidity coverage ratios, were designed with human behavior in mind.
They assume that deposit withdrawals occur gradually and are largely predictable. An AI‑driven mass exodus, however, could happen within minutes, outpacing the ability of regulators and banks to respond effectively.
Slok calls for a proactive approach, suggesting that policymakers need to consider new safeguards, such as real‑time monitoring of deposit flows, limits on automated transfer volumes, or requirements for AI agents to incorporate systemic risk considerations into their decision‑making algorithms. From a consumer perspective, the promise of higher returns is undeniably attractive, especially in a low‑interest-rate environment where the opportunity cost of holding cash is substantial.
Yet Slok warns that individual savers may overlook the broader systemic implications of their choices. He likens the situation to a classic coordination problem: while each household acting in its own best interest might be rational, the collective outcome could be detrimental to the stability of the financial system. To mitigate these risks, Slok proposes several practical steps. First, banks could enhance the attractiveness of their own checking and savings products by offering tiered interest rates that reward higher balances or longer holding periods.
Second, financial institutions might develop their own AI tools that provide customers with personalized recommendations while embedding safeguards against rapid, large‑scale transfers. Third, regulators could require transparency from AI‑driven financial platforms, mandating disclosures about how algorithms prioritize yield versus liquidity risk. In conclusion, Torsten Slok’s warning serves as a reminder that technological progress, while offering undeniable benefits, also introduces new forms of vulnerability. The automation of cash management through AI agents has the potential to improve individual financial outcomes, but it also carries the risk of amplifying systemic shocks if a large number of households simultaneously shift their deposits away from traditional banks.
By recognizing these dynamics early and implementing thoughtful policy and industry responses, the financial ecosystem can harness the advantages of AI without compromising the stability that underpins confidence in the banking system.