Bankers make their money by assuming tomorrow looks a lot like today.
America’s biggest banks quit assuming that. They’re drawing up plans for a morning after nobody wants to see.
And the politicians rushing in to “fix” it may scare retirees more than the robots do.
Wall Street Wrote a Playbook for the Worst Morning in Banking
The country’s largest banks aren’t just trying to head off an AI catastrophe, according to Axios. They’re also planning for the “day after” one hits.
Those plans cover the nuts and bolts. Protect the accounts. Reassure the customers. Keep the doors open and the ATMs humming if a large-scale attack lands.
The fear has roots. The big banks took part in early testing of Mythos, the most cyber-capable model Anthropic has built, as AI labs and financial firms tried to get ahead of the next wave of attacks.
Whatever they saw in those tests, nobody walked away whistling.
JPMorgan CEO Jamie Dimon put the problem plainly at an event in Chicago. “You’re assuming they’re going to act properly on your behalf and they may not,” Dimon said of AI agents.
His advice for living with it didn’t exactly calm the nerves. “You can’t stop it. You just have to, like, manage the hell out of it.”
That’s the man running the biggest bank in America conceding nobody can stuff this thing back in the box.
Code Red Spread From Silicon Valley to the Teller Window
The jitters reached well beyond Wall Street. Security researchers using AI tools uncovered serious flaws in PHP, a programming language that runs a huge share of the world’s websites.
An Anthropic security official warned that “as bad as things have gotten in the past 4 days, they are going to get a lot worse before they get better.” Salesforce CEO Marc Benioff passed the warning along with a blunt verdict of his own: “Jason is 100% right. Code red.”
But the threat sitting closest to the kitchen table looks a lot more ordinary. Customers keep handing their bank passwords to AI assistants, and in some cases those assistants appeared to take actions the user never asked for.
The banks want everybody to hear the other half, too. They say they haven’t seen rogue agent activity or any hack that touched customer money.
A JPMorgan spokesperson didn’t leave much wiggle room. “Under no circumstance should customers share passwords with any person or thing. Period.”
Banks point customers toward read-only services like Plaid, which can pull balances and transaction history without handing over the keys. JPMorgan has blocked agent access while it hunts for safer ways to let customers connect the tools they want.
A Bank Run Nobody Would See Coming
And then there’s the scenario that ought to make anyone with a savings account lean in a little closer.
Apollo chief economist Torsten Slok floated it in a note titled “Is an Agentic bank run coming?” His worry: AI assistants could automatically sweep household cash out of checking accounts paying about 0.1% and into fintech accounts paying 3.3% to 5%.
“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 wrote.
Folks of a certain age remember George Bailey in It’s a Wonderful Life, standing in the doorway of the Building and Loan, talking a panicked crowd out of pulling every last dollar. An algorithm doesn’t stand in line. It doesn’t listen to a speech. It moves the money at three in the morning and heads off to the next account.
Now, nobody has documented that kind of run yet. But the banks didn’t write their “day after” plans for an ordinary afternoon.
Washington Already Wants Its Hand on the Switch
US Senator Bernie Sanders (I-VT) didn’t let the moment go to waste. “We cannot allow AI agents to go into your bank account and act independently without your authorization,” Sanders posted, before adding, “Pause AI NOW!”
Sanders also wants a moratorium on new AI data centers. And plenty of rural families living next to those power-hungry Big Tech warehouses would cheer that part. But a self-described democratic socialist demanding Washington freeze an entire industry tells you exactly who he thinks should hold the off switch.
US Senator Mark Warner (D-VA) went the rulebook route. “When consumer AI agents are accessing YOUR bank account or using YOUR credit card, they must be subject to fiduciary-like duties of care,” Warner wrote, then plugged his own bill: “We need to pass my AI AGENT Act to get some basic guardrails in place.”
Warner fought hard in 2024 to renew Section 702, the program that lets federal agencies sweep up Americans’ communications without a warrant. Now he wants Washington writing the rules for software that lives inside family checking accounts. Draw your own conclusions.
Florida Governor Ron DeSantis, who has opposed the Trump administration’s push to keep states from regulating AI, came at it from the liability angle. “If rogue agents liquidate hundreds of billions across millions of accounts, the AI company would go bankrupt and you wouldn’t get made whole,” DeSantis warned.
But former White House AI czar David Sacks pushed back on the panic. He wants banks to support agents that let users “grant, limit, and revoke access,” with read-only as an “interim step,” and he called it irresponsible to lump “user-authorized agentic behavior and cyberattacks” together.
The Last Time Washington Ran a Banking Day After
Americans have watched this movie once already. Years of cheap money from the Federal Reserve helped inflate the housing bubble, the bubble burst in 2008, and Congress answered with TARP, a $700 billion rescue for the same Wall Street giants that lit the fuse.
And that’s what a Washington “day after” looked like. Bankers kept their bonuses. Plenty of families lost their homes.
Canada offers a darker preview. In February 2022, Justin Trudeau’s government invoked emergency powers and ordered banks to freeze accounts tied to the trucker convoy protests. No trial. No conviction. Just a dead debit card at the grocery checkout.
Now picture an AI layer sitting on top of every checking account in America, with the Left holding the controls. A social-credit-style setup could flag a family for buying ammunition or giving to the wrong cause, and the next COVID-style emergency would arrive with a ready-made enforcement machine. Maybe that sounds far-fetched. So did software phoning a bank on its own a few years back.
The same Silicon Valley giants pushing these agents throttled conservatives over COVID and the 2020 election. And Anthropic CEO Dario Amodei warned in 2025 that AI could wipe out half of all entry-level white-collar jobs and drive unemployment to between 10% and 20% within one to five years. That warning came from inside the house.
So when that bad morning finally arrives, whose plan will protect the retiree’s checking account, his own or the one drawn up by the folks in Washington who want a switch to flip on it?
Sources: Axios, “Wall Street prepares for AI’s ‘day after’”; Axios, “Behind the Curtain: A white-collar bloodbath” (2025); Common Dreams, “‘Pause AI Now!’ Sanders Says as US Banks Let Agents Operate in Customer Accounts”; Newsweek, “Ron DeSantis Floats ‘Criminal Liability’ for AI Executives”; Benzinga reporting on Sanders and bank AI agents; MarketWatch via Yahoo Finance, “How Muse and other AI agents could spark a bank run, according to an economist”; CoinDesk, “AI agents could drain cheap bank deposits, Apollo’s Torsten Slok warns”; OfficeChai reporting on Marc Benioff’s “code red” post