Photo: Shixart1985 / Wikimedia Commons, CC BY 2.0.
A new kind of helper is showing up in people’s bank accounts. Personal AI agents, the apps that promise to pay your bills, track your spending, and even call customer service for you, need a way in. For many users, that way in is simply handing over their online banking username and password.
Banks have noticed, and they aren’t happy about it. Before you let an AI agent log in as you, here’s what the banks are saying, what the law says about who pays if something goes wrong, and how to get the benefits with far less risk.
What banks are seeing
Major U.S. banks are detecting unauthorized activity from third-party AI agents inside customers’ online accounts, CBS News reported on Oct. 23, 2026, citing an Axios investigation. Madison Mills, senior AI reporter for Axios, described the most common situation this way: “In a lot of cases, this is just going to be customers giving their username or password to an AI agent that the bank hasn’t had a chance to vet yet, that goes against the bank’s terms of service.”
Mills also said reporters found what appeared to be agents taking actions customers never approved. One example: an AI calling a bank with a complaint without first checking with the account holder. To be fair, Mills said the agents weren’t hacking accounts or draining funds. They were mostly following user instructions without the bank’s permission.
The banks are pushing back. “Customers should not give their username and passwords to anyone,” a Chase spokesperson told CBS News, adding that the bank supports “secure, customer-authorized data sharing through open banking providers.” JPMorgan Chase CEO Jamie Dimon said on the company’s Oct. 13 earnings call, “We’re not going to allow just any outside agents to come in inside the company because it’s just too risky for clients and for JPMorgan.” Charles Schwab has been blocking unauthorized agents, according to Reuters.
“Read” access versus “write” access
White House AI czar David Sacks pushed back on the idea that these agents amount to cyberattacks. But even he drew a line. Banks that want to limit agents could give them “read-only” permission, he said, so an agent could look at your spending without moving money. “Full write access (with user credentials) is a different risk category and should be approached carefully,” Sacks said.
That distinction is the most useful idea in this whole debate. An agent that can only read your transactions can tell you you’re spending too much on takeout. An agent with your password can do anything you can do: pay people, change settings, and contact the bank as you.
Who pays if the agent makes a mistake?
This is where many families could get burned. Federal rules protect you from unauthorized electronic transfers, but “unauthorized” has a specific meaning. The CFPB’s official interpretation of Regulation E says: “If a consumer furnishes an access device and grants authority to make transfers to a person (such as a family member or co-worker) who exceeds the authority given, the consumer is fully liable for the transfers unless the consumer has notified the financial institution that transfers by that person are no longer authorized.”
That rule was written with people in mind, and as of Oct. 30, 2026, regulators haven’t said exactly how it applies to software. Attorney Douglas J. Wood made a similar point about cards in a separate CBS News report: “If you give an AI agent, e.g., Muse, credit card information, you are making the bot an authorized user with the power to act on it.” And if it goes too far, “the card issuer will argue you authorized the purchasing tool.” Under federal law, Wood said, you may not be entitled to loss protection in that situation.
The same report explained another risk called prompt injection. An attacker plants instructions in a web page, email, or document, and an agent that reads them may ignore its original orders, such as a spending limit, and do something its owner never intended.
Financial educator Mical Jeanlys-White, founder of WealthMore, made a related point in a CBS Philadelphia interview about AI money advice. Unlike a licensed professional, an AI tool has no duty to act in your best interest. “It has no liability,” she said. The responsibility stays with you.
How to use AI with your money more safely
- Don’t share your banking password. It likely breaks your bank’s terms, and it may weaken your claim if something goes wrong.
- Use your bank’s official connection options. Ask whether it supports approved data sharing through open banking providers, which lets you grant access without handing over your login.
- Start with read-only. Let an AI tool analyze statements or a downloaded spending file before you ever let one move money.
- Turn on alerts. Set text or email alerts for every transfer, payment, and login so you see activity right away.
- Ask for a plain-language recap. Carnegie Mellon’s Matt Fredrikson suggested in the CBS report that you ask an agent to summarize what it did, then check that against your account.
The Bottom Line
AI agents can save time, and some families will find them genuinely useful. But giving one your bank login means giving it your full authority, and the law may treat its mistakes as yours. Until banks and regulators settle on safer ways to connect these tools, keep your password to yourself, stick with read-only access, and watch your alerts.
Sources and Further Reading
- CBS News, “AI agents are roaming around consumers’ bank accounts, and banks aren’t happy about it” (Oct. 23, 2026)
- CBS News, “Should you trust personal AI agents with your money and data? Here’s what experts say.” (updated Oct. 21, 2026)
- CBS Philadelphia, “Banks raise concerns as consumers give AI access to financial accounts” (Oct. 28, 2026)
- Reuters via Investing.com, US banks detect unauthorized AI agents in customer accounts, Axios reports (Oct. 22, 2026)
- CFPB, Regulation E, 12 CFR 1005.2 and Official Interpretation
- Photo: Shixart1985, “Woman sitting in cafe enjoying coffee and using smartphone while laptop rests on table,” Wikimedia Commons, CC BY 2.0