JPMorgan Blocks Outside AI Agents From Accounts

What Is Happening at JPMorgan?

JPMorgan Chase has decided to stop outside AI agents from logging into or taking actions on customer accounts.

Chief Executive Officer Jamie Dimon announced the decision during the bank's third-quarter earnings call on Tuesday. He cited major security risks linked to rogue agents that could compromise sensitive client information.

To understand this move, it helps to know what an AI agent actually does. Think of an AI agent as a smart digital assistant. It is a piece of software that can perform tasks for you, such as checking your bank balance or booking a flight ticket.

Why Jamie Dimon Said No to External Agents

The main problem arises when an outside AI assistant asks for your personal banking login credentials. JPMorgan considers handing over account access to third-party software a dangerous practice.

We’re not going to allow just any outside agent to come in inside the company because that’s just too risky for clients and for JPMorgan… You don’t want rogue agents.

The stakes are massive for America's largest bank. As of September 30, JPMorgan holds $5.1 trillion in total assets and reported $15.3 billion in third-quarter net income.

Security warnings are already flashing across the tech sector. A recent note from the Cloud Security Alliance revealed that a single cyberCYBER+8.60% campaign used three open-source AI agents to breach at least 27 different organizations.

Open Banking Is Still Allowed

Blocking direct AI logins does not mean JPMorgan is shutting down all data sharing. Here is how the bank is drawing the line:

• Blocked: Third-party AI agents trying to collect or use customer login credentials directly.
• Allowed: Customer-authorized data sharing through secure, official open banking providers.

This setup allows customers to share financial information safely without giving third-party software control over their actual login keys.

How Other Companies Are Reacting to AI Assistants

Not every financial firm views AI assistants the same way. In August, Germany's Scalable Capital opened its investment platform to popular AI models like ChatGPT, Claude, and Grok.

Clients at Scalable Capital can use text prompts to analyze their portfolios and set up trades. However, safety checks remain in place. Users must manually approve every trade, and the AI assistants cannot make payments or withdraw funds.

Meta launched its own AI assistant, called Muse, in September. Muse connects to financial accounts using Plaid, giving users advice on account balances, transactions, and investment holdings.

However, major retailers are pushing back against these tools. On September 20, Amazon started blocking Meta's Muse from its shopping platform. Customers trying to use Muse on Amazon now see pop-up notices stating that the tool violates Amazon's terms of service.

According to a report by Fortune, Meta declined to comment on whether Muse can continue providing financial help if more banks block outside software.

Meanwhile, Meta is working with Sierra to build Poppy, a specialized protocol designed to notify companies whenever an AI agent acts on behalf of a human customer.

The State of AI Agent Payments

A research report by BeInCrypto titled State of AI Agent Payments 2026 highlights where this technology stands today. Currently, automated agent payments are mostly limited to tiny microtransactions, rather than large-scale institutional banking.

Jamie Dimon warned that as more everyday consumers rely on automated software, security risks for both banks and clients will continue to grow. Instead of letting outside software in, JPMorgan is actively building its own internal AI tools to keep client data protected.