In a recent commentary, Torsten Slok, the chief economist at Apollo Global Management, sounded an alarm about a subtle yet potentially destabilising force emerging from the rapid advancement of artificial intelligence: the rise of autonomous AI agents capable of managing personal finances. While the promise of AI‑driven financial advisors, robo‑investment platforms, and smart budgeting tools has been widely celebrated for its convenience and efficiency, Slok points out that these same technologies could inadvertently trigger a modern‑day bank run. The mechanism he describes is straightforward: AI agents, programmed to optimise returns for their users, will continuously scan the market for the best yield on cash holdings.
When they detect that a traditional checking account offers a near‑zero interest rate, they will automatically move funds into higher‑yield products such as money‑market funds, short‑term Treasury bills, or even newly emerging crypto‑based savings solutions. This systematic reallocation, performed at scale across millions of households, could drain the low‑interest deposits that banks rely on for liquidity and funding. Slok emphasizes that the danger does not stem from a sudden panic or a coordinated human protest, but rather from the cumulative effect of countless algorithmic decisions made in real time.
Each AI agent acts independently, following a set of optimisation rules that prioritize the financial well‑being of its user. However, when thousands or millions of such agents receive the same market signal—a modest uptick in the yield of a money‑market fund, for example—they will all execute similar transactions almost simultaneously.
The result is a rapid outflow of cash from checking accounts, which traditionally serve as a cheap source of funding for banks. Unlike a classic bank run, where depositors line up at the teller window out of fear, this AI‑driven scenario could unfold silently, behind the scenes of digital banking platforms, and without any overt sign of panic among consumers. The implications for the banking sector are significant. Banks have long depended on a stable base of low‑interest deposits to fund loans, invest in securities, and meet regulatory capital requirements.
These deposits are typically considered “sticky” because most consumers keep a portion of their cash in checking accounts for everyday transactions, even if the interest earned is negligible. If AI agents systematically divert these funds to higher‑yield alternatives, banks could face a shortfall in the cheap, stable funding they need to operate. In response, banks might be forced to raise deposit rates, tighten credit standards, or seek more expensive wholesale funding, all of which could compress profit margins and potentially lead to tighter credit conditions for borrowers. Moreover, the speed at which AI agents can act amplifies the risk.
Traditional bank runs were limited by the physical act of withdrawing cash or writing checks, which took time and often required a visible crowd of anxious depositors. Today’s digital infrastructure allows transactions to be executed in milliseconds.
An AI‑driven shift of funds can happen in seconds, leaving banks with little time to react or to communicate reassuring messages to their customers. The rapidity also complicates the role of central banks and regulators, who may find it challenging to monitor and mitigate such swift liquidity drains using conventional supervisory tools. Slok suggests several potential mitigations.
First, banks could develop their own AI‑powered tools that not only compete with third‑party agents but also incorporate loyalty incentives, such as tiered interest rates that increase with the length of deposit tenure. By offering a more attractive, dynamic yield structure, banks could retain a larger share of household cash. Second, regulators might consider imposing disclosure requirements on AI financial advisors, ensuring that users are aware of the broader systemic impact of mass fund migrations. Transparency could help temper the speed of automated reallocations, especially if agents are required to factor in systemic risk metrics alongside individual return optimisation.
Another avenue is the creation of a “deposit stability buffer,” a regulatory measure that would require banks to hold a certain proportion of high‑quality liquid assets proportionate to the volatility of their deposit base. If banks anticipate that AI agents could cause sudden outflows, they could pre‑emptively bolster their liquidity cushions, thereby reducing the chance of a funding crisis.
Additionally, collaborative industry initiatives could establish shared standards for AI agents, encouraging them to incorporate a modest “stay‑put” bias for a portion of deposits, especially in accounts that serve essential transaction functions. Beyond the immediate banking concerns, Slok highlights the broader societal implications. The convenience of AI agents may encourage households to keep lower balances in checking accounts, potentially reducing the availability of cash for everyday transactions and increasing reliance on digital payment ecosystems.
While this shift aligns with the overall trend toward a cash‑less economy, it also raises questions about financial inclusion, especially for demographics that are less comfortable with fully automated financial management. In conclusion, Torsten Slok’s warning serves as a timely reminder that technological progress, while beneficial, can generate unintended systemic risks. The advent of autonomous AI agents capable of optimizing personal cash holdings introduces a novel pathway for liquidity strain in the banking system—one that operates quietly, swiftly, and at scale.
Stakeholders across the financial ecosystem, from banks and fintech firms to regulators and policymakers, need to anticipate these dynamics and develop proactive strategies. By fostering collaboration, enhancing transparency, and designing incentives that align individual optimisation with systemic stability, the industry can harness the advantages of AI without compromising the resilience of the financial system.