How Muse and other AI agents could spark a bank run, according to an economist

An Apollo Global Management chief economist has raised a scenario worth understanding: agentic AI—systems that act autonomously on a user's behalf—could, in theory, contribute to a bank run. The argument is structural rather than predictive. If AI agents can monitor news, account balances, and market signals continuously, and move funds instantly without waiting for a human to log in and decide, the speed at which deposits could leave a bank may rise meaningfully.

That matters because bank runs are largely a coordination problem. Deposits are typically stable when customers believe others will stay put. Tools that let many depositors act faster, and in closer unison, could compress the window banks have to respond—even if no single agent is acting irrationally.

For markets, the takeaway is not a forecast but a lens: watch how quickly deposit flows can now move, how regulators and banks adapt their liquidity and monitoring frameworks, and whether AI-driven transactions receive any new guardrails. The risk is less about AI having opinions than about how fast it can execute them.