In a recent commentary, Torsten Slok, chief economist at Apollo Global Management, sounded the alarm on a potential new threat to the stability of the banking system: the rise of sophisticated artificial‑intelligence agents that can automatically reallocate personal cash holdings. While the concept of a bank run is traditionally associated with panic‑driven withdrawals triggered by rumors or financial distress, Slok suggests that a more subtle, technology‑driven mechanism could emerge. In this scenario, AI‑powered financial assistants—already embedded in many consumer apps and platforms—would continuously scan the market for the most attractive yields on short‑term, low‑risk instruments.
When they detect that a checking account is offering a rate significantly below what is available elsewhere, these agents could instantly move funds into higher‑yielding alternatives such as money‑market funds, short‑term Treasury bills, or even emerging digital‑asset products. The core of Slok’s warning lies in the speed and scale at which such automated reallocation could occur. Unlike a traditional bank run, which typically unfolds over hours or days as depositors line up at teller windows, an AI‑driven shift could happen in seconds, triggered by a single data point—say, a modest increase in the interest rate offered by a competitor. If millions of households rely on the same AI algorithms, the collective impact could be massive.
The result would be a rapid outflow of deposits from banks that have traditionally relied on low‑interest checking accounts as a cheap source of funding. These deposits are essential because they enable banks to lend to businesses and consumers at a spread that underpins profitability.
Why might AI agents behave this way? The answer is rooted in the design of many modern personal‑finance tools.
These applications promise users the best possible return on their idle cash, often employing machine‑learning models that compare rates across a broad spectrum of financial products. The promise of “optimizing your money” is a powerful selling point, and users have come to expect that their digital assistants will act in their best financial interest without requiring manual intervention. As a result, developers have built in automated execution capabilities—sometimes referred to as “auto‑invest” or “auto‑sweep” features—that can move money at the push of a button, or even without any user interaction once certain thresholds are met.
When such features are combined with real‑time data feeds, the system becomes capable of reacting to even minor market fluctuations. For example, if a large online bank raises its checking‑account APY from 0.01% to 0.10%, an AI agent monitoring that change could instantly transfer a portion of a user’s balance to capture the higher rate. Multiply that by the millions of users who have enabled automatic sweeping, and the cumulative effect could be a sudden, sizable reduction in a traditional bank’s deposit base. The implications for banks are significant.
First, a sharp decline in low‑cost deposits forces banks to seek more expensive funding sources, such as wholesale borrowing or issuing higher‑interest certificates of deposit, which compresses net interest margins. Second, the perception of a bank’s vulnerability can become self‑fulfilling: if market participants notice a bank’s deposit base eroding, they may demand higher rates or withdraw funds themselves, amplifying the stress.
Third, regulators could view rapid, technology‑driven outflows as a systemic risk, prompting new oversight measures or capital‑requirement adjustments. Slok also points out that the risk is not limited to traditional banks.
Any institution that relies heavily on retail deposits—credit unions, community banks, and even fintech firms offering cash‑management services—could be exposed. Moreover, the emergence of decentralized finance (DeFi) platforms, which often promise even higher yields on short‑term cash equivalents, adds another layer of competition. AI agents that are programmed to chase the highest return may start routing funds to these alternative ecosystems, further draining conventional banking channels.
Mitigating this risk will likely require a multi‑pronged approach. Banks could enhance the attractiveness of their checking accounts by offering tiered interest rates, fee waivers, or value‑added services that are difficult for AI agents to quantify. Transparency around how AI‑driven products operate and the inclusion of safeguards—such as user‑approved thresholds before any transfer is executed—could also help. From a regulatory perspective, supervisors may need to monitor the concentration of automated fund‑movement capabilities across the financial system and consider stress‑testing scenarios that incorporate rapid AI‑induced deposit shifts.
In conclusion, while the notion of AI agents sparking a bank run may sound futuristic, the underlying mechanics are already in place. The convergence of real‑time data, automated execution, and a consumer appetite for higher yields creates a perfect storm that could destabilize the traditional deposit‑funding model. Slok’s warning serves as a reminder that financial innovation, however beneficial, must be balanced with prudential oversight to ensure that the speed and efficiency of technology do not inadvertently undermine the stability of the banking sector.