Should You Let an AI Agent Into Your Bank Account? What JPMorgan's Block Means (October 2026)

Not with your password, if you bank with JPMorgan Chase. On October 13, 2026, CEO Jamie Dimon said the bank will not let outside AI agents such as Meta's Muse log in to customer accounts and act on them, and he warned that customers who hand an agent their login take on the risk themselves: "There's a liability transfer." Read-only connections through approved data-sharing services still work. The bigger lesson applies at any bank: give an AI agent the narrowest access that gets the job done, and never the keys.

What JPMorgan said

The comments came on the bank's third-quarter earnings call. Banking Dive and Fortune quoted Dimon:

• "We're not going to allow just any outside agent to come in inside the company," because it is "too risky for the clients and for JPMorgan."
• "You don't want rogue agents, and there could be thousands of them."
• "If people are using them, they should understand the risk. There's a liability transfer."

A JPMorgan spokesperson told Fortune the bank does not allow third-party agents that obtain customers' banking credentials to access accounts, and that it supports "secure, customer-authorized data sharing through open banking providers." CNBC named two agents affected: Meta's Muse and the personal agent from Instinct, a startup that raised $1 billion in September. Dimon also said the bank uses AI agents internally and is building tools to protect customers from outside ones.

What still works, and what doesn't

The block is about how an agent gets in, not about AI as such.

• Blocked: an outside agent logging in to your Chase account with your username and password, then acting on it. CNBC reported that this includes sending money and paying bills, and that the bank gave no timeline for safeguards that would let outside agents do this.
• Still allowed: customer-authorized data sharing through open banking providers. CNBC reported that agents can still read some Chase account data through connectors such as Plaid. Fortune reported that Muse's finance features already connect this way.

In practice, an agent can likely still tell you what you spent on travel last quarter. It cannot log in as you and pay the bill.

Why "liability transfer" is the phrase to remember

Most people assume that if money leaves their account without their say-so, the bank will treat it as fraud. Dimon's point is that an agent you gave your password to is, from the bank's view, acting with your authority. The Information reported that the bank's position is that a customer who hands login details to an agent that takes unwanted actions is responsible.

How that plays out in a real dispute depends on your account agreement and the consumer-protection rules that apply to you. We have not seen a test case yet. But the safe assumption today is the one Dimon spelled out: if you shared the password, the mistake may be yours to absorb.

This isn't only a banking story

Three weeks before the JPMorgan call, Amazon began blocking Muse from shopping on Amazon.com. Bloomberg reported that the block started on September 20, 2026, and Amazon's spokesperson said agents that buy on a customer's behalf "should operate openly and respect service provider decisions about whether or not to participate." Meta said Muse cannot see users' passwords or payment information.

The pattern is the same in both cases. Personal agents are arriving that book, buy, and pay. The companies on the other end want those agents to identify themselves and come through doors the company controls, not log in as you. Expect more of this. Fortune reported that Meta declined to say whether Muse could still offer financial help if more banks follow JPMorgan.

Not every bank is taking the same line yet. Wells Fargo's CEO said on the same day that the bank will "evaluate how third-party models and agents can provide value-added capabilities to our clients." As of October 15, no other major US bank had announced a block like JPMorgan's.

What to do if you use an AI agent

1. Check what your agent has. If you typed a banking username and password into any AI agent, change that password now. Then reconnect, if at all, through the bank's own data-sharing option, which usually appears as a connection screen hosted by your bank.
2. Separate reading from acting. Read-only access for spending summaries or cash-flow questions is a reasonable trade. Anything that can move money, pay bills, or change settings should be a tool your bank or employer has approved, with you confirming each step.
3. Keep work and client money out of it. Never give a personal agent access to a business account, a client trust account, or an employer card. That is a policy and compliance question before it is a convenience one.
4. Read the terms of both sides. Your agent's terms say what it does with your credentials. Your bank's terms say who pays when something goes wrong. If they don't line up, the bank's terms usually win.

What happens next

JPMorgan says it is building tools to handle outside agents safely but has not given a date. Meta is working with Sierra on a protocol for how agents identify themselves and get permission, Fortune reported, which is the kind of standard banks and retailers say they want. We'll update this page if JPMorgan changes its policy or other banks follow.

---

Sources

• Fortune, 'You don't want rogue agents': Jamie Dimon bars outside AI agents like Meta's Muse from JPMorgan accounts (October 15, 2026)
• Banking Dive, JPMorgan CEO Dimon warns customers of 'liability transfer' with AI (October 13, 2026)
• CNBC, JPMorgan Chase blocks outside AI agents like Muse, Instinct from directly accessing accounts (October 13, 2026)
• The Information, JPMorgan Blocks External AI Agents From Accessing Client Accounts (October 2026)
• Banking Dive, What Wells Fargo, Citi and Goldman Sachs' CEOs are saying about AI (October 15, 2026)
• Bloomberg via The Star, Amazon blocks Meta's Muse AI agent from its retail site (September 22, 2026)