In a recent commentary, Torsten Slok, the chief economist at Apollo Global Management, sounded the alarm on a new, technology‑driven risk to the banking sector: the emergence of sophisticated artificial‑intelligence agents that can move consumer funds at unprecedented speed and scale. While AI has been celebrated for its ability to streamline financial planning, automate transactions, and enhance customer service, Slok warns that the same capabilities could be weaponized—intentionally or inadvertently—to trigger a modern‑day bank run.
At the heart of the concern is the way AI agents are being programmed to optimize personal finances. These algorithms continuously scan a user’s portfolio, compare interest rates across a multitude of accounts, and execute transfers to the highest‑yielding vehicle available.
In a world where checking accounts often pay near‑zero interest, it is logical for a well‑designed AI to recommend moving cash into money‑market funds, short‑term Treasury bills, or high‑interest savings accounts that promise better returns. The problem arises when the decision‑making process is fully automated and the AI’s recommendations are acted upon without human deliberation.
Imagine a scenario in which a popular personal‑finance app integrates an AI assistant that automatically rebalances a user’s cash holdings each night. The assistant monitors rates offered by traditional banks, fintech platforms, and even emerging decentralized finance (DeFi) protocols.
If a new, higher‑yield product appears, the AI instantly transfers the user’s idle dollars from a low‑interest checking account to the new offering. For a single user, this is a harmless convenience.
However, when millions of users employ the same algorithmic logic, the cumulative effect can be dramatic. Slok points out that the speed and uniformity of these transfers could create a feedback loop reminiscent of classic bank runs, but without the panic‑driven stampede that historically characterized such events. Instead of depositors rushing to the teller windows out of fear, AI agents would systematically withdraw funds whenever a more attractive rate emerges.
The result would be a rapid, large‑scale outflow of deposits from traditional banks, especially those that have relied on low‑interest checking accounts to fund loan portfolios and other assets. The implications for banks are twofold. First, a sudden depletion of deposits would force institutions to sell assets—often at unfavorable prices—to meet liquidity needs, potentially eroding capital buffers.
Second, the perception of vulnerability could amplify the problem: if market participants observe a bank’s deposit base shrinking, they may adjust their risk assessments, leading to higher borrowing costs and tighter credit conditions. In extreme cases, the combination of actual cash outflows and deteriorating market confidence could push a bank toward insolvency. Regulators have traditionally focused on mitigating bank runs through deposit insurance, liquidity coverage ratios, and supervisory oversight. Slok suggests that these tools may be insufficient in an environment where the trigger is algorithmic rather than emotional.
He advocates for a proactive approach that includes: 1. **Transparency Requirements** – Financial institutions should disclose the proportion of deposits that are subject to automated, AI‑driven transfers. This data would help supervisors gauge systemic exposure. 2.
**Rate‑Stability Mechanisms** – Banks could offer modestly higher rates on checking accounts or introduce tiered interest structures that reward longer‑term balances, reducing the incentive for rapid churn. 3. **AI Governance Standards** – Fintech firms and banks that deploy AI agents need clear governance frameworks that limit automatic withdrawals to predefined thresholds or require explicit user confirmation for large moves. 4.
**Stress‑Testing Scenarios** – Central banks should incorporate AI‑induced deposit flight into their stress‑testing models, evaluating how quickly a bank’s liquidity position could deteriorate under a coordinated algorithmic outflow. Beyond regulatory measures, Slok emphasizes the role of consumer education.
While many users appreciate the convenience of “set‑and‑forget” financial tools, they must understand that the algorithms are acting on market conditions that can shift rapidly. A modest increase in a money‑market fund’s yield today could be offset by a sudden drop in that fund’s credit quality tomorrow. By encouraging users to periodically review the logic behind AI recommendations, the industry can mitigate blind reliance on automated decisions.
The broader financial ecosystem is also evolving. The rise of decentralized finance platforms, where smart contracts execute yield‑optimizing strategies without human oversight, adds another layer of complexity. If AI agents begin to route funds into these blockchain‑based products, traditional banks could face competition not only for deposits but for the very infrastructure that underpins modern payments.
In summary, Torsten Slok’s warning is a call to recognize that the same technology designed to make personal finance more efficient can inadvertently create a new systemic risk. The potential for AI agents to orchestrate massive, synchronized withdrawals challenges the conventional wisdom that bank runs are purely a psychological phenomenon. By combining regulatory foresight, prudent product design, robust AI governance, and informed consumer behavior, the financial industry can harness the benefits of artificial intelligence while safeguarding the stability of the banking system.