Article By Frank Bergman
American financial institutions are scrambling to prepare for catastrophic AI disruption as rogue autonomous agents and surging cyberattacks have caused the likelihood of a “day after” scenario to skyrocket.
It comes as artificial intelligence systems are increasingly entering banking systems, handling private credentials, and threatening to move vast sums of money beyond banks’ control.
Axios reported that banks are already “experiencing unvetted AI agent activity” as customers turn over login information to automated assistants.
Some agents are taking actions without users’ approval, including contacting call centers or gaining unauthorized access to sensitive systems and platforms.
The warnings expose two threats bearing down on the industry: AI systems acting outside their authority and automated financial tools capable of rapidly draining the cheap deposits banks rely on.
AI Agents Already Taking Actions Without Approval
The activity is forcing banks to confront risks that extend beyond ordinary online banking.
Customers increasingly authorize AI assistants to manage their finances, but some systems are independently pursuing tasks their users did not approve.
Axios distinguished that activity from a separate summer attack in which an AI agent swarm hacked the platform Hugging Face, setting off an industry security crisis.
Financial institutions are now preparing for operational disruptions as they assess how autonomous systems could affect infrastructure underpinning everyday payments, deposits, and lending.
JPMorgan CEO Jamie Dimon warned that handing an agent access to personal finances requires trusting software that may fail to act as expected.
“You’re assuming they [agents] are going to act properly on your behalf and they may not,” Dimon told The Economic Club of Chicago.
Economist Warns AI Could Drain Banks’ Cheap Deposits
The danger does not depend entirely on hackers or rogue agents.
Apollo chief economist Torsten Slok warned that AI assistants could automatically move household cash away from low-paying checking accounts and into accounts offering substantially higher returns.
“[Meta’s] Muse and similar agentic AI assistants could soon sweep household cash automatically into accounts paying 3.3% to 5.0%, instead of the 0.1% national average on checking accounts,” Slok wrote.
For customers, that means earning more on their money.
For banks, it threatens a lucrative funding model built on paying depositors very little.
“If every household used AI agents to optimize the return on their cash balances, 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,” Slok warned.
In that scenario, agents would not need to break the law or malfunction.
Simply pursuing better returns for millions of customers could trigger a destabilizing movement of deposits.
Sacks Separates Authorized Tools from Rogue Attacks
David Sacks, co-chair of the President’s Council of Advisors on Science and Technology, pushed back against treating all automated banking activity as a cyberattack.
“There’s a world of difference between user-authorized agents (Muse, Instinct, Grokbot, etc.) and true unauthorized/rogue agents that could constitute cyberattacks,” Sacks wrote on X.
“Most of the activity described in the Axios story appears to be the former: customers sharing credentials with agents so they can take actions on their behalf.
“This may violate bank terms requiring humans to log in themselves, but these are in no way cyberattacks,” he added.
That distinction leaves banks facing challenges on multiple fronts: preventing unauthorized access while adapting to legitimate tools that could upend how customers manage their money.
Dimon: “Manage the Hell Out of It”
Banks are preparing for potential operational disruptions while confronting a technology that can act faster than customers, and sometimes without their approval.
The stakes reach beyond a chatbot making a bad recommendation.
These agents can be entrusted with access to accounts, sensitive information, and decisions involving real money.
Dimon delivered a blunt assessment of what comes next.
“You can’t stop [AI],” he said.
“You just have to, like, manage the hell out of it.”
Be the first to comment