Unvetted AI agents are already operating inside U.S. bank accounts. “Keep a human in the loop” is the right instinct, but it can’t be the whole defense.

Last week, the fraud conversation picked up a new character: software that logs in as your member.

Axios reported on Oct. 22 that major U.S. banks are detecting unauthorized third-party AI agents operating inside customer accounts. According to its sources, some of those agents appear to act independently of what the user asked them to do. Some banks have tracked agents contacting call centers. And the hardest part, the banks said, is telling whether an agent is doing what the customer wanted or got in without authorization at all.

This isn’t a forecast. It’s happening now, at the largest institutions in the country.

The biggest banks are slamming the door

The response from the top of the industry has been fast and blunt.

Charles Schwab is blocking outside AI agents from client accounts, including agents that use clients’ own login credentials or browser automation, American Banker reported. “We are not going to allow that,” CEO Rick Wurster said. He told Bloomberg Television that Schwab gets a weekly report on these blocks and stops large agents “in the tens and twenties each week,” according to The Next Web. His core worry is liability: “We can’t control it, the client can’t control it, something could go wrong and who’s liable?”

JPMorgan Chase, Citigroup, Bank of America and Schwab have all said AI agents can’t move customer money, PYMNTS reported. Jamie Dimon warned of “rogue agents.” Bank of America’s Brian Moynihan said a human should stay in the loop to approve money leaving an account. PNC’s William Demchak called agents transacting inside accounts “incredibly unsafe,” adding: “For that to ultimately take place, there’s going to have to be laws passed on liability shifting.”

Per Axios, Morgan Stanley has begun blocking third-party agents, and JPMorgan, Bank of America and Citi are restricting them. Most banks are moving toward read-only access, the same way budgeting apps can view balances without moving money. Banks are also preparing “day after” plans to secure or restore customer access quickly if a breach occurs. US and European regulators are already questioning banks about AI agents, The Next Web noted.

Why this is a fraud problem, not just a policy problem

Here’s the uncomfortable part. An agent using a member’s real username and password doesn’t look like an intruder at login. The credential is valid. The device may be familiar. Every gate built around “is this the right login?” says yes.

Policy can say agents may only read. Enforcing it is harder, because, as the banks themselves told Axios, it is difficult to tell a user-directed agent from one that has gained access without authorization. Chase is telling customers not to share usernames and passwords with anyone. Good advice. Not every member will follow it.

That means the real decision point moves from the login to the transaction. What matters is what happens after the session starts: the transfer that doesn’t fit, the new payee at 2 a.m., the call to the contact center from something that isn’t quite a person.

Real-time detection isn’t the same as real-time protection

The industry’s instinct is right: someone has to approve money leaving an account. The question is who, and how fast.

Schwab can staff a weekly review of agent blocks. Most credit unions can’t put a fraud analyst on every session, every night, every weekend. An alert that waits in a queue until Monday is a record of the loss, not a defense against it.

Agents don’t work business hours. They don’t get tired. They run at machine speed. So the response has to as well.

The old model: Detect → Alert → Wait for a human.

The model this moment demands: Detect → Assess → Act.

Where RembrandtAi® fits

RembrandtAi® doesn’t stop at telling you something looks wrong. It ingests events across every channel (card, ACH, remote deposit, online banking, ATM), scores each one in real time with multi-stage AI, including neural networks, statistical analysis and machine learning, and then acts directly in your banking core. Member alerts, card restrictions, transaction blocks, account freezes and access suspension fire automatically, 24/7/365, without waiting on staff to read an alert.

It doesn’t matter whether the session was opened by a member, a fraudster with stolen credentials, or an AI agent that went off-script. RembrandtAi® judges what’s actually happening in the account, as it happens. When behavior breaks from the member’s pattern, the countermeasure is already underway.

Humans still matter. Your team sets the thresholds, works the exceptions and owns the member relationship. RembrandtAi® makes sure nothing waits on them at 2 a.m.

The approach has a track record. Launch Credit Union adopted RembrandtAi® at the end of 2020 and saw a 97% reduction in fraudulent transactions, saving $1.8 million in 2021, American Banker reported. As CFO Kevin LeBeau put it: “If you’re uploading three hours later, that money is already gone.”

The takeaway

The biggest banks in America just told the world they’re finding software inside customer accounts that they didn’t invite and can’t always identify. Their answer is to block, restrict and keep a human in the loop.

Credit unions need the same protection, without the 24-hour war room.

Identity is a moment. Risk is continuous.

See how RembrandtAi® acts in real time, in your core. Request a demo.

Sources

1. Madison Mills, “Scoop: Unvetted AI agents are accessing U.S. bank accounts,” Axios (via Yahoo Finance), Oct. 22, 2026. https://finance.yahoo.com/technology/ai/articles/scoop-unvetted-ai-agents-accessing-090006442.html
2. PYMNTS, “Banks Crack Down on Unauthorized AI Agents Operating Inside Customer Accounts,” PYMNTS, Oct. 22, 2026. https://www.pymnts.com/news/banking/2026/banks-crack-down-on-unauthorized-ai-agents-operating-inside-customer-accounts/
3. Cristian Dina, “Schwab and JPMorgan are blocking AI agents from moving customer money,” The Next Web, Oct. 16, 2026. https://thenextweb.com/news/schwab-blocks-ai-agents-meta-muse
4. Dan Shaw, “Schwab blocks unauthorized AI agents, points clients to Charley,” American Banker, Oct. 15, 2026. https://www.americanbanker.com/news/schwab-blocks-unauthorized-ai-agents-points-clients-to-charley
5. “Florida credit union overhauls its fraud prevention with big assist from AI,” American Banker, May 6, 2022. https://www.americanbanker.com/news/ai-revolutionized-how-this-credit-union-prevents-fraud