Major US Banks Are Blocking AI Agents That Log In to Customer Accounts With Their Credentials

Major US banks are detecting unauthorized third party AI agents operating inside customer accounts, and some of those agents have tried to take actions that appear independent of user instructions, Axios reported on October 22, 2026. Morgan Stanley has begun blocking third party agents from its platforms, and JPMorgan Chase, Bank of America and Citi are also restricting them, according to the report.

The agents at issue are AI agents that customers hand their own bank username and password so the software can log in and act for them, something most banks' rules forbid even when the customer agrees. An AI agent is software that carries out multistep tasks, such as paying a bill or moving money, with little human input. Reuters followed the same day with two people familiar with the situation, who said major banks have been blocking third party agents from operating inside accounts and making transactions.

The report comes nine days after JPMorgan Chase CEO Jamie Dimon told analysts on the bank's October 13 earnings call that it would not let "just any" outside agent inside, warning of "rogue agents" that "could be thousands of them" (CNBC).

What did Axios report about AI agents in bank accounts?

Axios reported on October 22, 2026 that major US banks are finding third party AI agents inside customer accounts without bank approval, usually because the customer shared their login credentials with the agent. In other cases, banks tracked agents that contacted call centers or appeared to act on their own, people familiar with the issue told Axios.

Bank executives described the situation as horrifying, according to the report, and one official said the industry has grave concerns about agents taking steps that could affect customers' finances. Axios named Morgan Stanley, Bank of America, JPMorgan and Wells Fargo among the banks racing to contain the risk. Some banks are also running "day after" scenario planning so they can quickly secure or restore customer access if an agent causes harm, sources said.

Which banks are blocking AI agents, and what do they still allow?

Morgan Stanley, JPMorgan Chase, Citi and Charles Schwab have each said, or been reported, to block third party agents from acting inside accounts, while most banks are moving toward read only access that lets an agent see account data without taking any action, according to Axios.

A Morgan Stanley spokesperson told Axios the bank hasn't observed instances of "rogue AI agents accessing our platforms." Schwab CEO Rick Wurster said earlier in October that the firm blocks agents to keep control over authentication, and that blocks are rare.

Why are banks worried about agents that use customer logins?

Many banks cannot reliably tell whether activity in an account is directed by the customer, by an agent the customer approved, or by an agent that gained access on its own, sources told Axios. Executives fear what an agent with login access might do once a user has given it purchase authority or the ability to move funds.

Liability is the other open problem. When customers share their login ID and password with an agent, banks typically flag and block those transactions, but the transactions cannot be deemed unauthorized because the customer handed over the credentials, a person in the banking industry told Reuters. Dimon made the same point on October 13: people using outside agents "should understand there's a liability transfer," he said (CNBC). Engadget reads the federal Electronic Fund Transfer Act the same way, arguing that a customer who gives an agent access would be unlikely to get money back if the agent made a mistake.

How does this compare with how retailers treat AI agents?

Retail is split, and Walmart has taken the opposite line from the banks: on October 21, 2026, one day before the Axios report, it announced an Agent Gateway that lets approved agents search and buy on its site, according to TechCrunch. Its Agent Profiles, due in November, let customers register each agent in their account, see its purchases separately and either approve every transaction or set a spending limit.

The banks' statements and Walmart's plan point the same way on one issue: both want agents to come in through a channel the company controls rather than with a customer's password. JPMorgan says it supports "customer-authorized data sharing through open banking providers" (CNBC), and Citi offers read only access through approved providers (Retail Banker International). The difference is that Walmart's channel lets an agent spend money within limits, while the banks' channels stop at reading data. Amazon blocks AI agents on its store, according to TechCrunch, and Delta Air Lines and Yelp have restricted agents on their own sites, The Verge reported on October 18.

What we don't know yet

• Which banks saw agents act beyond user instructions, and in how many accounts. Axios did not name them, and Morgan Stanley says it has seen no rogue agents.
• Which agent products were involved in the incidents Axios described.
• When any of these banks will let agents take actions such as payments. JPMorgan said there is no timeline (CNBC).
• What Wells Fargo is doing about agent access.
• Whether regulators will set rules on how customers authorize agents and who is liable for their mistakes.