In a recent commentary, Torsten Slok, a senior economist at Apollo Global Management, sounded the alarm on a new, technology‑driven risk to the banking sector: the potential for artificial‑intelligence‑powered agents to systematically drain cheap deposits from traditional checking accounts. While the idea may sound like something out of a futuristic thriller, the underlying mechanics are rooted in very real financial incentives and the accelerating capabilities of autonomous software. At its core, the concern revolves around the way AI agents can be programmed to continuously monitor the interest rates offered by a wide array of financial products. When a higher‑yield alternative—such as a money‑market fund, a short‑term bond ETF, or a newly launched high‑interest savings account—appears, the AI can instantly execute a transfer of funds from a low‑rate checking account to the more attractive option.
In a world where billions of dollars sit idle in accounts that pay near‑zero interest, even a modest improvement in yield can represent a significant gain for households. An AI that can act on those opportunities at scale, without human hesitation, could move money far more quickly and in far larger volumes than any individual depositor could manage.
Slok points out that this behavior is not merely theoretical. The financial industry is already seeing the emergence of “robo‑advisors” and algorithmic investment platforms that automatically rebalance portfolios, harvest tax losses, and chase yield.
Extending these capabilities to everyday cash management is a logical next step. Imagine a personal finance assistant that, after scanning a user’s bank statements, identifies that the checking account is earning 0.01% while a competing institution offers 0.50% on a comparable product. The AI could then initiate a transfer, notify the user, and repeat the process whenever a better rate surfaces. When millions of users employ similar agents, the aggregate effect could be a rapid outflow of deposits from traditional banks.
Why does this matter to banks? Cheap deposits—money that banks can hold at low cost—are a cornerstone of their business model. They fund loans, securities, and other higher‑margin activities.
When a large share of these deposits disappears, banks may be forced to raise the interest they pay on checking accounts, tighten credit standards, or sell assets to meet liquidity requirements. In extreme cases, a sudden, coordinated shift of funds could resemble a classic bank run, albeit one driven by algorithmic decisions rather than panic. The risk is amplified by several trends. First, the proliferation of open‑banking APIs gives third‑party developers unprecedented access to account data and transaction capabilities.
This openness, while fostering innovation, also lowers the technical barriers for building sophisticated AI agents that can move money on a user’s behalf. Second, the competitive landscape of consumer banking is evolving. Fintech firms and challenger banks routinely advertise higher yields on savings products, using digital channels to attract depositors who are increasingly price‑sensitive. Third, the macro‑economic environment—characterized by low baseline rates but occasional spikes in short‑term yields—creates frequent arbitrage opportunities that AI agents can exploit.
Regulators are beginning to take note. In the United States, the Federal Reserve and the Office of the Comptroller of the Currency have issued guidance on the safe use of fintech APIs, emphasizing the need for robust authentication, monitoring, and risk‑management frameworks. In Europe, the Revised Payment Services Directive (PSD2) already mandates strong customer authentication and imposes strict limits on third‑party access. However, the speed at which AI agents can act may outpace existing supervisory tools, which are traditionally focused on human‑driven behavior and slower, batch‑processed transactions.
To mitigate the threat, banks could adopt several strategies. Enhancing the attractiveness of their own deposit products—through tiered interest rates, rewards, or bundled services—can reduce the incentive for customers to chase higher yields elsewhere.
Investing in AI themselves to provide a counter‑service that offers personalized cash‑management advice while retaining funds within the institution is another avenue. Moreover, improving real‑time monitoring of large, rapid outflows can help banks detect early signs of an algorithm‑driven run and respond proactively, perhaps by adjusting liquidity buffers or communicating directly with customers.
From a consumer perspective, the rise of AI cash‑management agents presents both opportunities and risks. On the one hand, users stand to benefit from higher returns on idle cash without having to manually track rates across dozens of platforms. On the other hand, the automation could lead to unintended consequences, such as reduced cash availability for everyday expenses, increased exposure to less‑stable financial products, or loss of control over where funds are held. Transparency about the decision‑making criteria of these agents, as well as clear opt‑out mechanisms, will be essential to maintain trust.
In summary, Torsten Slok’s warning highlights a subtle but potentially disruptive force shaping the future of banking: AI agents that automatically seek out the best yield for household cash. While the technology promises efficiency and better returns for consumers, it also threatens to erode the cheap deposit base that underpins traditional banks, possibly triggering liquidity strains reminiscent of a bank run.
Stakeholders—including banks, regulators, fintech developers, and consumers—must collaborate to design safeguards, promote responsible AI deployment, and ensure that the benefits of automation are realized without compromising financial stability. The conversation is just beginning, and the industry’s response will determine whether AI becomes a catalyst for innovation or a source of systemic risk.