Economist Warns AI Could Trigger a Bank Run as Agents Optimise Returns on Cash

AI tools could one day move billions of pounds out of low-paying current accounts without a hint of panic in a branch queue, an economist has warned. Instead of fear-fuelled withdrawals, software could quietly sweep household cash towards higher interest rates, transforming how banks fund themselves.

That is the scenario outlined by Torsten Slok, chief economist at Apollo, the alternative-asset manager, in a note published on Sunday, 27 September 2026.

Slok said AI agents could eventually sweep household cash from low-yield checking accounts into products paying substantially more. He compared the 0.1% average US checking rate with rates of 3.3% to 5% available from some alternative accounts.

The scenario does not describe a bank run happening now. Slok's argument is that AI agents could automate the rate-shopping and account-switching decisions that currently require consumers to compare products and move their money themselves.

AI Could Make Deposits Less Sticky

Slok wrote that if every household used AI agents to optimise the return on its cash balances, banks could lose a large share of the cheap deposits they rely on to make loans, which he said could become a problem for the wider financial system.

The scenario differs from a conventional bank run driven by fear and rapid withdrawals. Households would be responding to differences in yields, with software potentially identifying and acting on those differences automatically.

Banks that lose low-cost deposits could need to compete for replacement funding or pay more to retain existing customers. The cost of that funding is one factor in how banks price loans and manage their balance sheets.

A $10,000 balance earning 0.1% generates about $10 a year in interest. At 5%, the same balance would generate about $500, before tax and subject to the account's terms.

Not every customer would move money, and advertised rates can come with conditions, caps or other restrictions. Slok's scenario assumes widespread use of agents to optimise household cash balances.

New AI Agents Are Entering Personal Finance

New AI agents are increasingly being designed to perform tasks on behalf of users across digital services. Meta launched Muse on 8 September as a personal AI agent capable of performing tasks across apps and services. The company says Muse can act on a user's behalf, including opening a browser, filling in forms and negotiating with businesses.

Through a partnership with Plaid, Muse's finance experience can connect users' financial accounts and provide the agent with balances, transactions, investment holdings, mortgage information and other financial data. Plaid says consumers remain in control of which accounts they connect.

That does not establish that Muse currently sweeps money between bank accounts automatically. The public material shows access to financial information and financial guidance, not an autonomous system moving deposits between banks.

The existing financial-account connections nevertheless show how an AI agent can be given information needed to compare a household's financial position and products. Whether agents receive authority to execute transfers is a separate question.

Where the Money Goes Matters

The effect on banks would depend partly on where automated cash moves. If customers move from a low-paying bank to another bank offering more, the banking system still retains the money. The change would be the distribution of deposits between institutions and the price banks pay to attract and retain them.

If cash instead moves into money-market funds, Treasury securities or other assets outside bank deposits, banks would lose those deposits altogether. Slok's scenario therefore centres on automated decisions about household cash balances.

Instead of a customer leaving money in an account because changing banks takes time and effort, an agent could repeatedly compare available returns and act according to the user's instructions.

For banks, that would change how much customer inertia contributes to the stability of their deposit base.