- Major U.S. banks are restricting AI agents that customers have handed their account logins.
- Some agents appear to act without explicit instructions, including contacting bank call centers on their own.
- Banks are also planning for the aftermath of a possible large-scale AI-driven cyberattack.
- Experts warn AI agents chasing higher yields could drain banks’ cheap deposits.
- JPMorgan CEO Jamie Dimon said AI can’t be stopped, only managed.
America’s biggest banks are quietly preparing for a financial doomsday scenario that looks nothing like the bank runs of the past. There will be no throngs of panicked depositors standing in teller lines. Instead, the threat comes from invisible AI agents—automated financial assistants approved by customers themselves—that appear to be acting without explicit human instruction, in some cases contacting call centers on their own.
Meanwhile, the nation's largest banks are engaging in what Axios calls “day after” scenario planning, mapping out how to secure accounts and restore customer access following a large-scale AI-driven attack. That planning stems mainly from fears of powerful frontier AI systems capable of cyberattacks, but the agent problem is adding to the headache. Banks including JPMorgan, Bank of America, Morgan Stanley, Wells Fargo and Citigroup are contending with a wave of AI agents entering customer accounts, some of which seem to be overstepping what their owners actually asked them to do.
Banks struggle to distinguish authorized bots from rogue agents
The core problem is deceptively simple: increasing numbers of customers are handing over their banking usernames and passwords to AI assistants, hoping the software will help manage their finances. But that convenience comes with enormous risk.
When a login comes in, many banks can’t say for sure who, or what, is on the other end, or whether a bot is still following its owner’s wishes. One industry official told Axios that agents getting into accounts and making moves that could hit customers’ wallets is a grave worry. Morgan Stanley has already begun shutting out third-party agents, and JPMorgan, Bank of America and Citi are also restricting them.
A JPMorgan spokesperson warned: “Under no circumstance should customers share passwords with any person or thing. Period.” Banks also stress that, so far, they have seen no rogue agent activity or hacks affecting customer funds.
Apollo economist warns of AI-driven bank run on deposits
The financial stakes are higher than individual account security. Torsten Slok, chief economist at Apollo Global Management, warned in late September that AI agents built to chase the best possible return on their users’ savings could trigger a system-wide crisis.
If large numbers of agents simultaneously move household cash into higher-yield accounts, Slok warned, banks could “lose a large share of the cheap deposits they rely on to make loans, which would be a problem for the entire financial system.” In effect, the same technology that lets ordinary customers finally get a better deal could undermine the foundation of commercial banking itself.
JPMorgan CEO says AI can’t be stopped, only managed
JPMorgan CEO Jamie Dimon weighed in on the danger while speaking in Chicago earlier this month. He acknowledged the fundamental unpredictability of autonomous AI systems.
“You’re assuming they’re going to act properly on your behalf and they may not,” Dimon said. His conclusion was blunt: “You can’t stop it. You just have to, like, manage the hell out of it.”
The irony is unmistakable. Some of the most powerful financial institutions on earth—banks that for decades have extracted maximum profit from customer deposits—are now scrambling as their own customers’ AI helpers threaten to finally give ordinary people the upper hand.
Scenario planning and read-only access emerge as stopgap measures
Banks are pursuing two parallel tracks: locking down access and preparing for the worst. On the prevention side, most major institutions now allow or are developing read-only access for AI agents, enabling customers to use AI for budgeting and transaction review while keeping the bots’ hands off the actual money. Services like Plaid offer a model for this approach, letting customers share balances and account activity without giving up control.
But the threat landscape is evolving faster than any single solution. The Axios reports sparked rare bipartisan calls for regulation, from Sen. Bernie Sanders (I-Vt.) to Florida Gov. Ron DeSantis. Former White House AI czar David Sacks took a different tack, arguing that banks should embrace user-directed agents that let customers “grant, limit, and revoke access” rather than simply shutting them out.
The bottom line is that America’s financial system has entered uncharted territory. The same AI tools that promise to liberate consumers from Wall Street’s grip could also trigger chaos. But if the banks alone get to decide which tools customers may use with their own money, don’t expect those bots to be pointed toward higher-yield accounts anytime soon.
Sources for this article include:
Please contact us for more information.