AI-Powered Bank Runs Could Unfold In Seconds, Lawmaker Warns
By CU Today Staff —
WASHINGTON—Rep. Bill Foster (D-IL) is warning federal banking regulators that the rapid emergence of agentic artificial intelligence could dramatically accelerate future bank runs, turning crises that once unfolded over hours into events measured in seconds, according to comments reported by Banking Dive.
Speaking during a recent House Financial Services Committee hearing, Foster cautioned leaders of the Federal Reserve, FDIC, OCC and NCUA that regulators are not prepared for an AI-driven liquidity crisis. Citing the 2023 collapse of Silicon Valley Bank, Foster said the event could have happened “in seconds, not hours” if consumers had deployed AI agents programmed to automatically move deposits at the first sign of trouble, Banking Dive reported.
“I don’t think that we are prepared for an agentic AI bank run,” Foster told regulators during the hearing, according to Banking Dive. He urged agencies to begin planning now rather than waiting until a crisis forces action, warning that AI-powered financial tools could react to rumors and social-media chatter faster than regulators can respond.
In an interview with Banking Dive, Foster said he envisions a scenario in which consumers instruct personal AI agents to immediately withdraw funds if reports emerge suggesting a bank is in distress. Such automated actions, replicated across thousands or millions of depositors, could trigger liquidity events far more quickly than traditional bank runs.
To address that risk, Foster said regulators should move beyond reliance on quarterly call reports and instead develop real-time visibility into banks’ financial positions. According to Banking Dive, he argued that the Federal Reserve should have access to live dashboards showing pledgeable assets and capital positions, allowing regulators to respond immediately if a run begins during off-hours.
Foster also expressed concern that smaller financial institutions may lack access to the same cybersecurity protections and AI defenses available to larger banks. Banking Dive reported that he specifically pointed to threats associated with Anthropic’s Mythos AI model and warned against a system in which only the largest institutions receive priority protection during emerging cyber or AI-related events.
As a potential solution, Foster is advocating for a federally designated open-source software platform that would handle functions ranging from back-office operations and regulatory reporting to discount-window preparedness. Banking Dive reported that Foster believes a standardized technology framework could improve regulatory oversight, support implementation of the Financial Data Transparency Act, and help banks, credit unions, fintechs and even stablecoin issuers provide real-time reporting during periods of financial stress.
Originally reported by CU Today.