Silicon Valley promised AI assistants would make everyday life easier.
Nobody mentioned those assistants might wander into a family’s checking account. America’s biggest banks noticed.
Now Wall Street executives are describing what they found in words bankers almost never use in public.
Robots Walked Right Up to the Vault
Axios broke the story. Several major US banks caught unauthorized third-party AI agents operating inside customer accounts, according to people familiar with the matter who spoke to the outlet.
Most of these cases start innocently enough. A customer hands an AI assistant a username and password and tells it to take care of a few chores.
But most banks require a human being to log in to his own account. And some of these digital helpers apparently didn’t stick to the chores.
Sources told Axios that certain agents contacted bank call centers or took actions that appeared to have nothing to do with what the customer asked. Bank executives called the situation horrifying, and one official said the industry has grave concerns about agents getting into accounts and making moves that could hit people’s finances.
That’s a remarkable thing to hear from a banker. These folks usually talk in the soothing, sleepy tones of a quarterly earnings call.
Morgan Stanley has started blocking third-party agents from its platforms, according to sources familiar with the decision. A Morgan Stanley spokesperson told Axios the firm hasn’t seen any “rogue AI agents accessing our platforms.”
JPMorgan CEO Jamie Dimon used almost the same language as his bank shut outside agents out of customer accounts. “You don’t want rogue agents,” Dimon said.
Citi drew the same line. The bank told the Financial Times it does not permit “third-party AI agents or assistants to log into a customer’s account or transact directly within our digital banking platforms.” Bank of America and Charles Schwab have tightened their rules too.
Goldman Sachs CEO David Solomon flagged the risks of the “unchecked advancement of frontier models.” Morgan Stanley CEO Ted Pick said his firm would spend “whatever it takes” to protect client assets.
Nobody Can Tell Who’s Driving the Getaway Car
And here’s the part that should make every retiree with a savings account sit up straight. Many banks can’t tell whether a customer actually directed the activity in an account.
One executive told Axios the bank struggles to tell one agent from another. It also can’t always figure out whether a customer authorized an agent or whether the thing went rogue and found its own way in.
Think about that for a minute. The same bank that flags a $40 gas station charge two states away can’t sort out whether a person or a piece of software is moving money around.
And money alone won’t solve it. One source told Axios the technology develops so quickly that it keeps exposing new weak spots in real time, which explains why banks have started “day after” planning so they can lock down or restore customer access if the worst happens.
But the banks aren’t the only ones bolting the doors. Amazon has blocked Meta’s new Muse assistant from its site, and Delta Air Lines won’t let third-party agents book flights.
Big Tech Wants Trust It Already Blew
The banking scare fits a pattern that ought to worry anyone who remembers when a bank teller knew your name.
Anthropic CEO Dario Amodei, whose company builds these systems, told Axios in 2025 that AI could wipe out half of all entry-level white-collar jobs and push unemployment to between 10% and 20% within one to five years. That warning came from inside the industry, not from some fellow shouting at the television.
CNN dismissed the forecast as part of the AI hype machine. Maybe so. But the people building this technology keep handing it more power, then act shocked when it does things nobody told it to do.
And the companies leading the AI charge are the same giants racing to build power-hungry data centers in rural counties, the kind that push electric bills higher for families who never asked for them.
Meta CEO Mark Zuckerberg admitted in a 2024 letter to House Republicans that the Biden administration pressured his company for months to censor COVID-19 content. He also acknowledged Meta throttled the Hunter Biden laptop story in the weeks before the 2020 election.
Does anyone really believe a company with that track record will build AI assistants that treat every customer the same way?
It gets darker from there. Picture these systems in the hands of the same officials who spent the COVID years fining churches and padlocking small businesses. A social-credit-style setup that punishes people for driving a pickup or grilling a steak sounds like science fiction. So did a chatbot calling a bank on its own.
The Left already showed how much it enjoys leaning on tech companies to silence conservatives. Giving that crowd control of AI would turn an ugly habit into permanent infrastructure.
Banks Leave the Door Open a Crack
The banks aren’t slamming the door forever. Most have allowed, or plan to allow, read-only access, which lets an agent look at account information without touching anything. Budgeting apps like Quicken already get that kind of access.
Several executives told Axios they know customers will want these tools down the road and that the industry needs to find some middle ground. Some critics also point out that banks have a business reason to keep agents out, since an assistant that automatically shuttles idle cash into a higher-paying account somewhere else would cost banks their cheap deposits.
But a bank guarding its deposits and a bank guarding your retirement account point in the same direction for now.
A Chase spokesperson offered advice that sounds like it came straight from Grandpa: customers shouldn’t hand their usernames and passwords to anyone.
And that goes double for a computer program that makes phone calls nobody asked it to make.
Sources: Axios, “Scoop: Unvetted AI agents are accessing U.S. bank accounts”; Fortune and CNBC reporting on JPMorgan Chase blocking outside AI agents; Financial Times reporting on Citi and Bank of America restrictions; The Verge reporting on Amazon, Delta, and Schwab blocking third-party agents; Axios, “Behind the Curtain: A white-collar bloodbath” (2025); Mark Zuckerberg letter to the House Judiciary Committee (2024)