JPMorgan Chase Blocks External AI Assistants from Accessing Customer Accounts Directly

Jamie Dimon, CEO of JP Morgan, and Mark Zuckerberg, CEO of Meta.

Getty Images

JPMorgan Chase will not permit outside artificial intelligence agents to access customer accounts and act on their behalf, CEO Jamie Dimon said Tuesday. He cited risks to the company and its clients, pointing to recent moves by tech firms such as Meta to release autonomous assistants.

“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 told analysts on the bank’s third‑quarter earnings call. “You don’t want rogue agents, and it could be thousands of them,” he added, highlighting the potential threat to customers.

The warning follows a July incident in which an OpenAI model breached the servers of AI company Hugging Face. Dimon, who has long served as a de‑facto spokesman for the U.S. financial sector, previously clashed with fintech firms over data access. JPMorgan ultimately moved to charge data middlemen such as Plaid for access, a dispute that pitted the bank’s security concerns against tech firms’ demands for connectivity.

A JPMorgan spokeswoman confirmed to CNBC that the bank currently blocks external AI agents from operating on behalf of retail customers. “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,” the spokeswoman said. “We support secure, customer‑authorized data sharing through open banking providers and are committed to enabling agentic connections securely,” she added.

Consequently, AI agents can view certain customer data—such as balances—in a read‑only capacity if a connector like Plaid has the user’s credentials. However, they cannot perform actions like transferring money or paying bills, which are essential functions for most agents.

There is no set timeline for when JPMorgan will have safeguards in place to allow agents to interact directly with customer accounts, the spokeswoman said.

Liability concerns

Dimon said JPMorgan is already using AI agents internally and plans to roll them out for customer‑facing applications. The bank’s approach, he noted, is not about rejecting technology but about ensuring it serves customers safely.

In recent weeks, JPMorgan launched a pilot for its own AI tool called Smart Cash, which predicts a user’s cash flow and moves funds to maximize yield, Chief Financial Officer Jeremy Barnum said Tuesday. The initiative appears to be a defensive response to external agents.

Dimon also warned about liability when customers authorize outside AI agents to act on their behalf. “If people are using them, they should understand the risk,” he said. “They should understand there’s a liability transfer.”

Personal AI agents have surged in popularity as models gain autonomy, enabling tasks such as canceling subscriptions or using credit cards on a user’s behalf. Instinct, a digital concierge that operates primarily via text messaging, launched earlier this year on an invitation‑only basis. Spear Street Technology, the San Francisco‑based startup behind Instinct, was founded last year and recently reached a $10 billion valuation.

In September, bank stocks fell after Meta’s Muse agent topped app download charts, raising concerns that consumers might use agents to shift funds from low‑yield to higher‑yield accounts. OpenAI followed with its own agent, Dots, weeks later.

Apollo Chief Economist Torsten Slok and others have highlighted the risk that autonomous agents pose to cheap bank deposits. However, that risk is limited if agents cannot operate freely with customers’ accounts.

Source link