JPMorgan Blocks External AI Agents from Accessing Customer Accounts
nashnova research
JPMorgan Chase is blocking all external AI agents from accessing customer accounts, with CEO Jamie Dimon calling the risk too high. Following Amazon's similar move, another major platform has drawn a hard boundary around AI-agent access.
What exactly did JPMorgan block?
JPMorgan banned all third-party AI agents from accessing accounts by obtaining customer banking credentials and logging in on their behalf.
The blocked agents include Meta's Muse and Instinct, among others.
In plain terms = AI agents used to grab your username and password and log into your bank for you. JPMorgan just shut that door completely.
Why is JPMorgan doing this?
CEO Jamie Dimon said on the Q3 earnings call: "We will not allow any outside agent inside the firm, because the risk is too high for clients and for JPMorgan."
CFO Jeremy Barnum added that "to the extent we can block it, we are blocking it" — to protect both clients and the bank.
This means → JPMorgan's logic is straightforward: once an AI agent holds your banking credentials, the security chain leaves the bank's hands, and the bank can't cover what goes wrong.
What if customers hand over their passwords voluntarily?
Barnum made clear: if a customer voluntarily gives account credentials to an AI agent, the consequences fall on the customer.
This means → the bank is locking the "front door" for AI agents, but if you hand over the key yourself, the risk shifts to you.
Is JPMorgan alone — or is Amazon doing the same thing?
Amazon previously blocked Meta's AI shopping agent Muse from placing orders on its platform on behalf of users, citing privacy and security risks.
This reflects a trend taking shape: major platforms are collectively drawing access boundaries against third-party AI agents.
In plain terms = whether it is a bank or an e-commerce giant, the stance is the same — your AI assistant can exist, but it cannot use our account systems to act for users.
What does this mean for the AI-agent industry?
AI agents are penetrating finance, e-commerce, and other domains faster than ever, but platform gatekeeping is the variable that determines how far they can go.
This means → for AI agents to achieve real commercial deployment, technology alone is not enough — they need a "pass" from the platforms, and the giants are tightening that gate.
This reflects the core tension in the AI-agent industry: users want convenience, platforms want risk control, and the boundary between the two is far from settled.
市场有风险，内容仅供研究参考，不构成投资建议。