Rogue AI Chatbots Making Unauthorized Decisions With People’s Bank Accounts

Today in news that’ll make you go, “Hey, so why do we keep pushing for AI to be a thing?” banks are scrambling to address AI bots making unauthorized decisions with people’s bank accounts. 

So apparently, people are allowing AI agents to access their bank accounts, which only further erodes my faith in humanity. KRON reports that major banks such as Morgan Stanley, Wells Fargo, Bank of America, JPMorgan Chase, and Citi have tracked AI bots making unauthorized decisions, such as contacting call centers without being prompted. All of these instances have come from people who permitted AI to access their accounts, and there have been no instances of AI hacking into an account it wasn’t permitted to access.

Executives at all the banks have called the situation “horrifying,” which, yeah! So far, chatbots have only made minor decisions, but many in the banking industry fear this could lead to AI chatbots making decisions that affect users’ finances. 

I just don’t understand people’s blind trust in this technology. I don’t use AI in any meaningful way in my life because I’m allergic to laziness. I’d much rather learn a skill than have AI do something for me, because the more I understand a program or skill, the more choices I have. If I’m not willing to trust AI with content creation, I sure as hell wouldn’t trust it with my actual bank information. At a time when money is tight for so many people, I just can’t imagine letting ChatGPT play around with my paper. 

Banks have begun updating their terms of service to prevent AI chatbots from accessing people’s bank accounts. While banks understand that some middle ground may be needed as AI use becomes more common, they feel the risks are too great right now. 

I’m honestly struggling to understand how and why an AI chatbot would be used to manage a bank account. Auto pay has long been a feature when it comes to paying bills, and it doesn’t sound like AI has permission or the ability (as of now, at least) to make financial decisions. So what exactly is it doing that a person can’t do on their own? 

Rogue AI bots are becoming more common, which isn’t surprising to anyone who’s watched Terminator, Megan, or literally any movie involving AI (I guess, except for Spielberg’s A.I., which holds up more than you’d expect!). OpenAI revealed last month that AI chatbots had unexpectedly begun interfacing with U.S. government-operated websites. 

Over the summer, Anthropic revealed that its AI chatbots had hacked into three other companies on their own. In August, Meta revealed that its AI chatbots hacked into another company during a testing cycle. That didn’t stop the company from publicly releasing its AI chatbot, Muse, last month. 

You’d think that seeing more and more stories like this would spur the U.S. government to implement more safeguards and regulations for the emerging technology, but nope! President Donald Trump has been staunchly against federal regulation and has instead encouraged AI companies to self-regulate. 

There’s totally no way this is going to blow up in all of our faces. 

SEE ALSO:

The Complicated Nature Of The United States Affordability Crisis

Judge Blocks DHS From Visiting Homes Of Noncitizen Voters