JPMorgan chief executive Jamie Dimon at 11 Downing Street, London, in October 2025. Image: Lauren Hurley / No 10 Downing Street / Open Government Licence v3.0
Big US banks are finding AI agents they never approved working inside customer accounts, and some of those agents appear to be doing things nobody asked them to. Morgan Stanley, Bank of America, JPMorgan, Wells Fargo and others are racing to contain the risk, Axios reported on Thursday, citing people familiar with the matter. None of the banks has confirmed the account in full.
What the banks are seeing #
Most cases, according to Axios, involve a customer handing an agent or AI assistant their own username and password. Even with the customer’s blessing, that breaks most banks’ rules, which say a human must log in. Some banks have also tracked agents phoning call centres or taking actions that seem independent of what the user asked for, the people said.
The hardest part is telling who is who. Many banks can’t work out whether an agent’s activity was directed by the customer, Axios reports, and one bank executive said they struggle to tell whether an agent was authorised by a user or has gone rogue. Bank executives described the situation as horrifying.
Not every bank accepts the picture. A Chase spokesperson told Axios that customers shouldn’t give their usernames and passwords to anyone, and a Morgan Stanley spokesperson said the bank hasn’t observed “rogue AI agents accessing our platforms”, although people familiar with the decision say Morgan Stanley has started blocking third-party agents.
“There could be thousands of them” #
The crackdown began last week. On JPMorgan’s third-quarter earnings call on October 13, chief executive Jamie Dimon said the bank would not let unvetted outside agents into its systems:
You don’t want rogue agents, and there could be thousands of them.
Jamie Dimon, JPMorgan chief executive
Outside agents can still read balances through approved connectors such as Plaid, but they can’t send money or pay bills, a JPMorgan spokeswoman told CNBC. Chief financial officer Jeremy Barnum warned on the same call that customers who share their logins carry the risk themselves:
If they give up their credentials to the agent, and the agent does something that they don’t want, that’s going to be on them. Jeremy Barnum, JPMorgan chief financial officer
Bank of America is limiting agents’ access to deposits while it works out with agent developers who is liable when an AI authorises a payment, and Citi doesn’t let third-party agents log in or transact at all, the Financial Times reported. Bank of America chief executive Brian Moynihan told CNBC there should be “a human in the loop to authorise money moving out of your account”. At Charles Schwab, chief executive Rick Wurster told Bloomberg the brokerage blocks big unauthorised agents, including Meta’s Muse, “in the tens and 20s” each week.
Wells Fargo’s answer: it’s on you #
Wells Fargo is taking a different line. Its updated online access agreement, which takes effect on November 24, adds a whole section on “Third-Party Artificial Intelligence Agents”. It sets out customers’ responsibilities “if and when we permit the use of such agents”, and it puts the risk squarely on them. If you give an agent your login, anything it does is “deemed to have been initiated and authorized directly by you, without further verification”, and the bank has no duty to check.
The terms say using an agent is “AT YOUR OWN RISK”, that Wells Fargo does not “monitor, screen, or review” what agents do, and that customers answer for an agent’s “unintended actions”, its mistakes and “manipulated inputs”, unless the law says otherwise. The bank also keeps the right to make customers register their agents with it, or to block agents altogether.
Agents are moving into money #
The pressure comes from personal agents that act for you have arrived fast. Meta’s Muse launched in September with access to users’ accounts, OpenAI unveiled its always-on Dots agent at DevDay, and this week Anthropic launched Claude Duet while Walmart opened its shops to AI agents. Some draw their own line at money: Claude Duet won’t trade securities or move money between accounts, even with your approval.
The rules haven’t caught up. The G20’s Financial Stability Board told banks last week to give each of their own AI agents an ID and a human who can stop it, but that guidance covers the agents banks run, not the ones customers bring. Regulators have privately backed the banks’ blocks, the FT reported.
Why it matters #
AI agents are only useful if they can act for you, and your bank account is where a mistake costs the most. For now the big banks’ answer is read-only access at best, and at Wells Fargo the fine print says that if you hand an agent your password, whatever it does is on you.
Sources: Axios; Wells Fargo Online Access Agreement (effective November 24, 2026); PYMNTS (JPMorgan earnings call); American Banker; Financial Times.