Catenaa, Wednesday, November 04, 2026- AI agents handling financial transactions could create new banking risks and should remain under human oversight, JPMorgan said.
In a report published on November 2, the bank’s Global Technology Leadership Team said AI agents can increasingly access enterprise systems, retrieve information, and execute tasks with limited human involvement.
“Frontier technologies are moving from answering questions to taking action. Agents can increasingly interact with enterprise systems, access information and execute tasks with limited human involvement.” JPMorgan said.
JPMorgan cautioned that advances in AI capabilities should not automatically result in greater decision-making authority. Companies should establish clear operating restrictions and determine which systems and activities AI agents are permitted to access.
JPMorganChase recommended different levels of human supervision depending on the risks associated with an AI agent’s activities. It identified financial transactions, customer communications and changes to critical technology systems as activities requiring stronger safeguards than routine, reversible tasks.
Moreover, the bank said that agents should only be given the access required for the task and that limits should be enforced on the systems, data, and actions available to them.
“Make agents identifiable, monitor their actions in real time and preserve traceability for investigation and recovery,” JPMorgan said.
The bank said that enterprises should maintain the ability to restrict or stop an agent quickly and reassess boundaries as capabilities, use cases and dependencies change.
JPMorgan said frontier technology is compressing the time between vulnerability discovery, potential exploitation, and the need to respond, and that governance must move at the speed of the threat.
“Speed does not mean bypassing governance or accepting uncontrolled change. It is the product of governance designed before the crisis begins,” JPMorgan said.
