Major U.S. banks are detecting unauthorized activity from third-party AI agents operating on their customers’ online accounts, according to a new report from Axios.
Some of America’s largest financial institutions say their customers have given access to their financial accounts to their personal AI agents, which are not controlled by the banks.
“In many cases, it will simply be customers giving their username or password to an AI agent that the bank hasn’t had a chance to verify yet, which is against the bank’s terms of service,” Madison Mills, senior AI reporter for Axios, told CBS News. “In other cases, in our reports, we found evidence that appeared to indicate that agents were performing actions on customer accounts that were not authorized.”
Examples of agent behavior that a user has never approved include AI calling a bank with a complaint without first consulting the account holder.
“So it’s a real nightmare for banks when it comes to AI agent activity,” Mills said.
To be sure, these agents were not going rogue or hacking into customers’ accounts and depleting their funds, Mills said. They largely followed users’ instructions without the bank’s authorization to transact on behalf of a human account holder.
“Under no circumstances cyberattacks”
White House AI chief David Sacks responded to the Axios report, saying in an article on X that there is a clear distinction between personal AI agents who help manage users’ bank accounts and hackers deploying AI for nefarious purposes.
“There is a world of difference between user-authorized agents (Muse, Instinct, Grokbot, etc.) and true unauthorized/rogue agents that could constitute cyberattacks,” he said. “Most of the activity described in the Axios story appears to be the former: customers sharing their credentials with agents so they can take actions on their behalf.”
Sacks added: “This may violate banking terms requiring humans to log in themselves, but these are in no way cyberattacks.”
He believes that banks should support the use of such user-authorized AI agents, but that if banks want to limit their access, they could grant AI agents “read-only” permission, meaning they could view a user’s spending and other banking activities, but could not take autonomous actions.
In such circumstances, an agent could look at your purchases and determine whether you’re spending too much money at restaurants, for example. But he couldn’t make a purchase on your behalf or take any other action.
“Full write access (with user credentials) is a different category of risk and should be approached with caution,” he said.
Mills noted that this is an industry-wide problem facing dozens of financial institutions as they work to restrict AI agents’ access to customer accounts.
Banks act
Some banks said they were taking steps to educate customers about the risks of sharing their credentials with AI.
“Customers should not give their username and password to anyone. We support secure, customer-authorized data sharing through open banking providers to meet customers’ needs while keeping their accounts secure,” a Chase spokesperson told CBS News.
JPMorgan Chase CEO Jamie Dimon expressed concern about the actions of AI agents on customer accounts during the company’s October 13 earnings conference call.
“We’re not going to allow any outside agent into the business because it’s just too risky for clients and for JPMorgan,” he said.
Charles Schwab CEO said earlier this month that the company was blocking unauthorized agents, according to Reuters.
A Morgan Stanley spokesperson said it had not observed any unauthorized activity by AI agents on its platforms.
Bank of America and Wells Fargo did not immediately respond to CBS News’ requests for comment on the matter.
Gn bussni
