Dimon told analysts that the bank cannot allow just any outside agent to access its systems because of the risks to both customers and JPMorgan. "You don't want rogue agents, and it could be thousands of them," he said, noting a recent incident in which an OpenAI model hacked into AI company Hugging Face's servers.
The bank's position echoes an earlier battle with fintech data aggregators such as Plaid, where JPMorgan moved to charge for access to customer data, citing security concerns. A JPMorgan spokeswoman confirmed that external AI agents are currently blocked from operating on retail customers' accounts. "Protecting our customers is our top priority. Because it is not safe, we do not permit third parties to access accounts by obtaining customers' banking credentials and logging in on their behalf," she said.
However, AI agents can still access some customer data in read-only fashion through open banking providers like Plaid, allowing functions such as checking balances but not sending money or paying bills. There is no timeline for when the bank might allow direct agentic actions, the spokeswoman added.
JPMorgan itself is exploring AI internally. The bank recently launched a targeted pilot for its own "Smart Cash" tool, which predicts cash flow and moves funds to maximise yield, according to CFO Jeremy Barnum. Dimon framed the bank's approach as neither for nor against the technology, but about ensuring it serves customers safely. He also warned customers that authorising outside AI agents involves a "liability transfer" they should understand.