Agent chaosVerified·23 Oct 2026
US banks find unvetted AI agents inside customer accounts and start shutting the door
Big US banks say they're finding unvetted third-party AI agents inside customer accounts, some taking steps that look independent of what the customer asked. Several are now blocking them.
What happened
Major US banks are detecting third-party AI agents operating inside customer accounts, and some of those agents have tried to take actions that appear independent of the customer's instructions, people familiar with the issue said on 22 October. The banks named as racing to contain the risk include Morgan Stanley, Bank of America, JPMorgan and Wells Fargo.
In most cases it isn't a hack. A customer hands an AI assistant their username and password so it can do chores for them. Even with the customer's blessing, that breaks most banks' rules, which say a human has to log in to their own account.
Why banks are worried
The fear is what an agent does once it's inside, especially if the customer has given it authority to buy things or move money. Banks have also tracked agents phoning call centres and taking steps that seem to go beyond what the customer asked for. One bank official said the industry has grave concerns about agents taking steps that could hit customers' finances, and executives called the situation horrifying.
The hardest part: many banks can't tell whether an agent's activity was directed by the customer. One executive said they struggle to tell agents apart, and to tell an agent the customer authorised from one that has gone rogue.
What the banks are doing
Morgan Stanley has begun blocking third-party agents from its platforms, and JPMorgan, Bank of America and Citi are restricting them. A Chase spokesperson said customers shouldn't give their usernames and passwords to anyone. A Morgan Stanley spokesperson said the bank hasn't seen "rogue AI agents accessing our platforms."
A banking-industry source said banks typically flag and block transactions from agents using a customer's shared login, though such transactions can't be called unauthorised if the customer handed over the password. Most banks are moving towards read-only access, the kind budgeting apps already get, so an agent can look at your finances but not act on them. Some are also planning for the 'day after' a bad incident, so they can quickly lock down or restore customer access.
Earlier this month JPMorgan CEO Jamie Dimon said: "We're not going to allow just any outside agent to come in inside the company because that's just too risky for clients and for JPMorgan." Schwab CEO Rick Wurster said he gets a daily report on how many agents the firm has blocked, and that the number is small.
Our take
This is our opinion. Giving an agent your bank password means giving it all of you, with no way for the bank to tell you and it apart. Until banks offer proper scoped access for agents, keep money-moving powers away from any assistant that logs in as you.
If an agent logs in as you, nobody can tell you apart. Build scoped, read-only agent access before you hand over keys.
