JPMorgan Chase Bars Outside AI Agents From Direct Account Access

JPMorgan Chase is blocking outside AI agents, including tools called Muse and Instinct, from logging into and acting on customer accounts directly. The decision draws a firm line between the bank's systems and third-party software that tries to operate on a customer's behalf.

Chief executive Jamie Dimon has been vocal about the risks. He warned of rogue agents operating at scale, meaning autonomous software that could take many actions across many accounts faster than a person could monitor or stop them. The concern is less about a single bad transaction and more about how quickly a flawed or malicious agent could multiply its impact.

Blocking outside access is only part of the response. The bank says it is also building defensive tools of its own to detect and manage automated agents. The source material does not describe how those tools will work or when they will be deployed.

The move matters beyond one bank. A growing number of fintech and AI startups have offered products that log into people's accounts to aggregate data, move money or automate tasks, often using credentials the customer provided. If large banks refuse direct access, those services will have to rely on formal data-sharing channels, or the industry will need new rules for how agents identify themselves and what they are allowed to do.