JPMorgan Chase has barred third-party artificial intelligence agents from directly accessing retail customer accounts. CEO Jamie Dimon made the announcement during the bank’s third-quarter earnings call.
As personal AI agents rapidly gain popularity, large platforms and banks are growing increasingly concerned about both security risks and the potential loss of direct contact with users.
“We’re not going to allow just any… outside agent to come in inside the company, because it’s just too risky for clients and for JPMorgan,” Dimon said. “You don’t want rogue agents, and it could be thousands of them.”
A bank spokesperson confirmed the policy. JPMorgan does not permit third parties to access accounts by collecting customers’ login credentials and logging in on their behalf. The bank views this approach as creating unacceptable security risks.
At the same time, the organization supports secure data sharing through open banking providers. It has also said it is working to enable so-called “agentic connections” safely, though it has not given a timeline for when third-party agents might be allowed to interact directly with accounts.
For now, AI agents can obtain limited account information, such as balances, only through intermediaries like Plaid and strictly in view-only mode. They are prohibited from making transfers, paying bills, or carrying out any other active operations.
Dimon also addressed the question of customer responsibility, according to CNBC. People who use third-party AI tools, he said, need to clearly understand the risks involved.
“If people are using them, they should understand the risks. They should understand that there is a transfer of liability,” he emphasized. If data is misused, much of the responsibility falls on the user.
JPMorgan has by no means abandoned artificial intelligence. The bank is already using AI agents internally and is preparing to introduce them into customer-facing applications. Chief Financial Officer Jeremy Barnum announced the launch of a limited pilot of an in-house product called Smart Cash.
The system can forecast a user’s cash flows and automatically move funds toward higher-yielding options. Earlier this year, the bank’s chief analyst Derek Waldron outlined plans to deploy internal autonomous agents in 2026 capable of running complex workflows for several hours without human intervention.
Financial institutions are coming under growing pressure from technology companies seeking deeper access to customer accounts and data. JPMorgan has previously clashed with fintech firms over data access and began charging intermediaries for providing customer information.
If agents were able to move funds freely between accounts, low-yielding bank deposits could come under threat. The restrictions JPMorgan is introducing reduce that risk, but they also slow the development of external finance-focused AI services.
Against this backdrop, Amazon’s decision is particularly telling. The company has banned Meta’s new Muse AI agent from making purchases on Amazon.com. The move has further intensified the industry-wide struggle over who controls the relationship with customers when artificial intelligence begins acting on their behalf.
