Bank of America Warns Agentic AI Risks Banking Models

Bank of America issued a warning on Wednesday that agentic AI assistants, such as Meta Platforms Incorporated's Muse, present a multiyear evolutionary risk to traditional banking business models. Unlike standard chatbots, these autonomous agents can manage account administration and execute purchases independently.

Bank of America analyst Ebrahim H. Poonawala noted that these agents can identify excess liquidity and automatically shift funds into higher-yield investments. This capability threatens to drain trillions of dollars from low-interest checking and savings accounts, which would compress bank margins. The risk is already appearing in the market through the decline of financial sector ETFs like XLF and KBWB.

To counter this trend, the bank suggests that financial institutions may need to self-cannibalize by developing their own AI assistants. By integrating these tools into their own brokerage and wealth platforms, banks can attempt to keep customer assets within their own ecosystems.