BROKEN ARROW, Okla. — More than 1 in 4 Americans now use AI chatbots for personal finance questions, but new research from NerdWallet finds the results are far from guaranteed — and the risks are growing.
About 26% of Americans say they have used an AI chatbot to find answers to personal finance questions, according to NerdWallet. Of those who acted on that advice, 39% say it improved their financial situation, 29% say it hurt their finances, and 32% say it had no impact.
kjrh
“It’s not a slam dunk. If I ask AI a money question and take its advice, it might not work out for me,” Sara Rathner, senior credit cards expert at NerdWallet, said.
Part of the appeal is the format itself. Nearly 28% of AI users say they turned to chatbots because they wanted a non-judgmental response to their money questions.
“It kind of removes that element of feeling like somebody else has an opinion about your decisions that maybe is different from yours,” Rathner said. “You can prompt a chatbot not just to give you certain information, but even to respond to you in a certain way, to be more conversational, to be more formal. So it’s really this customized experience down to how it makes you feel.”
But that personalized feel comes with a significant caveat: AI chatbots can be agreeable to a fault.
“That’s a concern,” Rathner said. “You can certainly prompt it to be a little bit less agreeable, and to challenge your assumptions a little bit more. And that really comes back to the quality of responses you get from AI, in part, have to do with how you prompt it. And not everybody is as skilled in prompting AI as some, because it takes practice, and if you are not very experienced in using AI, you might get results that are not necessarily the right results for you.”
KJRH
Broken Arrow-area resident Aaron Carlson, who works in cybersecurity, said he sees value in the technology — with limits.
“I know everyone’s scared of it, but I feel like it’s a great tool and we should utilize it,” Carlson said.
But he also cautions users need to verify the information AI returns.
“You can’t always go off that — you shouldn’t let it do your homework for you, but it’s definitely a good tool,” Carlson said.
cathy
John Scukey, another local resident, said he uses AI as a starting point but does not stop there.
“I don’t go totally by what AI says because sometimes I’ll get, you know, contrary comments, so I’ll go and research it afterwards. But the AI is basically the first thing I go to,” Scukey said.
That approach aligns with what NerdWallet recommends. The quality of AI responses depends not just on how a question is asked, but on where the chatbot pulls its information.
“Is it pulling information from reputable websites, like government sources, or is it pulling from random websites that might have outdated information or even incorrect information?” Rathner said. “You really want to not just take the response and run with it, you want to follow up, ask more questions, and really fact check to make sure you’re getting accurate information.”
NerdWallet also found that about half of Americans — 49% — do not feel confident evaluating whether personal finance advice from an AI chatbot is accurate. And 1 in 5 people who followed AI financial advice acted on it immediately, without follow-up questions, additional research, or human input.
story blocks
AI agents can now spend your money
The stakes are rising beyond advice. AI agents can now be given access to bank accounts and credit cards directly, and in some cases make purchases without human input at each step.
“We’re beginning to see the foundation being set in the payment world, where with a human being’s consent, an AI agent can have access to payment methods, like a credit card number, and be given parameters, and then go out and shop on behalf of a human, and even pay for those items with the human being’s payment method of choice that they’ve provided to the AI bot,” Rathner said.
Major financial platforms are already moving in this direction. ChatGPT now connects to credit cards and bank accounts, reading real transaction data across more than 12,000 financial institutions. Robinhood’s agentic credit card gives AI its own virtual card number to make purchases on a user’s behalf. Visa and Mastercard have launched agentic commerce platforms that allow AI agents to compare prices and complete purchases without human input at each transaction.
Rathner said credit card companies have experience managing AI-related risk, but acknowledged the technology is still evolving.
“Just like your credit card company checks to see that you are behaving predictably and your purchases are typical for you, and if they’re not, they’ll send you notification that there might be fraud on your account,” Rathner said. “AI agents also have to behave in a predictable way, and once they don’t, there is a security risk that’s flagged.”
She added that the technology’s track record will ultimately determine how widely it is trusted.
“If it makes reservations — tries to make reservations at restaurants that haven’t been open for years, if it messes up the dates on your hotel booking so you arrive and there’s no hotel room for you — I mean, if that was a human travel agent, you’d fire them,” Rathner said.
The bottom line to protect your wallet and peace of mind when using AI for financial advice is simple.
“You really want to double and triple check the information that it’s giving you,” Rathner said.
This story was reported on-air by a journalist and has been converted to this platform with the assistance of AI. Our editorial team verifies all reporting on all platforms for fairness and accuracy.
Stay in touch with us anytime, anywhere —
