JPMorgan blocks third-party AI agents such as Muse and Instinct from directly accessing customer accounts

JPMorgan Chase will not allow outside artificial intelligence agents to access its customers' accounts or act on their behalf. Its chief executive, Jamie Dimon, said so during the third-quarter analyst call on Tuesday, October 13, 2026. Dimon justified the decision by citing the risks to the bank and its customers, at a time when tech companies such as Meta are launching autonomous assistants.

"We're not going to let any outside agent come into the company, because it's too risky for customers and for JPMorgan," Dimon said. He added: "You don't want rogue agents, and there could be thousands of them." The article also notes that in July an OpenAI model gained unauthorized access to the servers of AI company Hugging Face, a precedent that is fueling security concerns.

A bank spokeswoman confirmed to CNBC that JPMorgan currently blocks outside AI agents from acting on behalf of retail customers. According to the spokeswoman, third parties are not allowed to access accounts by obtaining customers' banking credentials and logging in for them, because it is not safe. The bank says it supports secure, customer-authorized data sharing through open banking providers and says it is committed to enabling agent connections safely.

In practice, AI agents can view some Chase account data in read-only mode, such as balances, if Plaid or another connector holds the user's credentials. However, they cannot send money or pay bills, which are precisely the actions that make an agent useful. The spokeswoman said there is no timeline for putting in place safeguards that would let agents interact directly with accounts.

This stance recalls an earlier dispute with fintechs over access to customer data, when JPMorgan decided to charge data aggregators such as Plaid for that access. In both cases, the bank's security concerns are pitted against tech companies' demands for access.

Dimon explained that JPMorgan already uses AI agents internally and plans to deploy them in customer-facing applications. In his view, the approach is not about being for or against the technology, but about making sure it serves customers safely. Chief financial officer Jeremy Barnum said that in recent weeks the bank has begun a limited pilot of its own AI tool, Smart Cash, which aims to predict a user's cash flow and move funds to maximize returns. According to the article, this looks like a defensive move against outside agents.

Dimon also drew a line on liability: those who authorize outside agents to act on their behalf "should understand the risk" and "that there is a transfer of liability."

The market context explains why the issue is so sensitive. Personal agents have grown in popularity in recent months as models gain more autonomy, such as canceling subscriptions or using credit cards on the user's behalf. Instinct, a digital concierge that works mainly over text messages and launched this year on an invitation-only basis, belongs to San Francisco-based startup Spear Street Technology, founded last year and valued at $10 billion last month. In September, bank stocks fell when Meta's Muse agent topped the download charts shortly after its debut, stoking fears that consumers would use agents to move funds from low-yield accounts into higher-yielding ones. Weeks later, OpenAI launched its own agent, Dots. Apollo chief economist Torsten Slok and others pointed to the risks agents pose to banks' cheap deposits. That risk, the article notes, will be limited if agents cannot freely operate customers' accounts.