Apollo's Slok warns AI agents may spark 'agentic bank run' over rate gap

Apollo Global Management's chief economist Torsten Slok has outlined a new systemic risk developing where artificial intelligence meets consumer banking, calling it a potential "agentic bank run." His warning revolves around a substantial interest rate disparity. According to Slok, AI assistants such as Muse could soon shift household deposits automatically into accounts offering 3.3 percent to 5.0 percent, far above the 0.1 percent that most US households now earn on checking accounts.

That rate gap is illustrated by a comparison of fintech products and FDIC averages. Neobank offerings from Adelfi, Sofi, LendingClub, LevelUp, Pibank, Revolut, Current, Fitness Bank, Varo, Wealthfront and AlumniFi show annual yields from 3.3 percent to 5.0 percent, while the FDIC national averages stand at 0.4 percent for savings and 0.1 percent for checking. That creates a spread reaching nearly five percentage points between the top fintech rate and typical bank checking returns.

Slok's argument is not that individuals seeking better returns pose a threat. Instead, he highlights the consequences if such behavior becomes automatic and ubiquitous. If agentic AI tools allow millions of households to optimise cash returns simultaneously, banks could face a swift, widespread loss of cheap deposits used for lending. Banks have historically counted on those deposits as "sticky" because most people do not actively chase small rate differences on transaction accounts. An AI managing the household's cash eliminates that inertia.

The framing as a potential bank run—rather than just a shift in deposit market share—underscores the systemic risk Slok emphasises: a rapid, coordinated, machine-led departure of deposits from banks could pressure the funding models that support lending. That could affect the broader financial system, not just individual institutions. As agentic AI tools grow more advanced and more commonly used by consumers, the rate gap Slok points to may continue serving as a marker for how real and how rapid this threat could become.

This is a scenario warning from Apollo's chief economist, not a live event, so today's direct market impact is limited. However, the underlying rate gap it highlights is genuine and already reflected in deposit flows. Slok's chart shows fintech and neobank products yielding 3.3 percent to 5 percent, versus FDIC national averages of 0.4 percent on savings and 0.1 percent on checking, a spread of up to roughly 4.9 percentage points. If agentic AI tools accelerate existing deposit migration by automating it at scale, the risk Slok flags is faster erosion of banks' cheap funding base, which would affect net interest margins, lending capacity and, under stress, funding stability. For now this is a forward-looking risk to watch rather than a confirmed trend, and regional bank commentary and deposit data should be monitored for early signs of the dynamic Slok describes.

Slok's concern is less about any single bank and more about what happens to the whole system if millions of households let an AI agent chase the best rate automatically, all at once.

• Apollo Global Management chief economist Torsten Slok has raised the prospect of an "agentic bank run", in which AI assistants automatically move household cash into higher-yielding accounts.
• Slok's note points to agentic AI tools potentially sweeping cash into accounts paying between 3.3 percent and 5.0 percent, compared with a national average of just 0.1 percent on checking accounts.
• An accompanying chart shows a range of fintech and neobank products, including Adelfi at 5.0 percent, Sofi at 4.5 percent, LendingClub and LevelUp at 4.2 percent, and Pibank at 4.1 percent, all well above FDIC national averages of 0.4 percent for savings accounts and 0.1 percent for checking accounts.
• Slok's concern is systemic: if households broadly used AI agents to optimise cash returns, banks could lose a large share of the low-cost deposits they depend on to fund lending.
• He frames this as a potential problem for the entire financial system rather than for any single institution.