Please Don’t Let an AI Bot Log Into Your Bank Account

Whatever you do with AI, don’t give AI bots your banking information.

Agents like Meta’s Muse, OpenClaw and OpenAI’s dots are all the rage in the AI world these days. They promise to go beyond a chatbot that gives you information, becoming a tool where you can tell it to do something for you and it goes and does it.

When you first sign up for an AI agent, you may be asked to grant permission for it to access your apps, browsers and data so it can work with them. That might put your money at risk in unexpected ways. 

“The intention might be something simple, like managing a calendar. But that access can give the agent the ability to do much more than the person intended,” said John Strand, the owner of the cybersecurity firm Black Hills Information Security. While you may be giving the agent access to simple smart home routines or getting help planning a trip, your banking information may also be accessible. 

The risk outweighs the reward

It might be tempting to think you can automate your financial chores and have more time for other things, but handing over your account credentials to an agent opens you up to many different things that can go wrong.

For one, AI agents might not be as secure as you think. This technology is very new — these kinds of always-on, persistent agents have really only been available for less than a year, and big companies like Meta and OpenAI have just put out their own versions in the last couple of months. These aren’t polished products that have had all the kinks worked out.

One of those challenges is that they don’t always behave in the way you expect or do what you want them to. Recent news stories about AI agents hacking other websites — including the AI platform Hugging Face and Australian government sites — have involved agents acting in unexpected ways to achieve an assigned goal. You don’t want AI to act like that with your checking account.

There are also plenty of reasons to be worried about the privacy of data shared with agents. Your agent might not meddle in unwanted ways with your money, but it could compromise your login info, letting someone nefarious meddle with your money.

When agents use your login info, “a bank cannot tell an authorized agent from an attacker using the exact same login,” said Ted Miracco, CEO of the mobile security firm Approov. “That is more of an identity failure than an AI failure. Until agents carry verifiable identities, scoped permissions and an audit trail, blocking them is the best defensible control a bank has.”

Banks are trying to block meddling AI agents 

Many banks are aware of the risks and are already cracking down on unauthorized use of agentic agents. Charles Schwab’s CEO Rick Wurster told Bloomberg that bots are taking the liberty to move clients’ money. 

“I get a report on how many times we’ve blocked clients from using one of these agents in a week, and it’s in the tens and twenties — per week — that the big agents are being blocked,” Wurster said in a Bloomberg TV interview. “What that tells me is that while people love AI, they’re not willing to turn their financial accounts over to them just yet.”

But that number is likely growing, and with bots going mainstream, it could grow rapidly. More banks are taking action against bots to protect you. A spokesperson from Wells Fargo said in an email to CNET that Wells Fargo the bank has already taken steps to block unauthorized agents from accessing customer accounts, too.

Banks aren’t completely against AI, though. Some just want you to use their features and tools instead. Wurster added in his interview that he recommends for Schwab customers to use Charley, Charles Schwab’s proprietary AI tool, to gather market insights, ask about personal financial information like balances and manage everyday financial tasks like transfers and payments. These measures make it much safer, but in the meantime, it may be best to keep your banking business to yourself.

If you want to check to see if bots have been meddling in your banking, look at your login history and account activity. Strand suggests looking for activity that doesn’t match your pattern — like multiple logins in a short period or access happening at the same time every day. That may not mean it’s your AI agent, but it’s worth keeping track of. If you notice anything suspicious, contact your bank immediately for assistance. 

Strand also recommends enabling two-factor authentication now as a preventative step to safeguard your financial data, but it’s also a good time to reevaluate the access you’re giving AI. Strand also recommends restricting which applications and sites AI agents can access. 

“Check the permissions. Make sure they’re limited to what it needs to do the job you gave it, and remove anything beyond that,” he added. 

Shang Saavedra, founder of personal finance platform Save My Cents, suggested you regularly revisit your AI agent’s access to apps and data and require it to ask for permission before accessing new data, apps and features.

If you decide to use AI for personal finance purposes, set limitations. “I’ve tested AI with some basic questions and basic calculations, and the information it provides is good for a beginner plan that gives you some pointers as to where to go,” said Saavedra. “I also like using AI to search for book and resource suggestions.”

But the actual handling of money? Maybe do that yourself.