JPMorgan CEO Jamie Dimon warns against “rogue” AI agents calling customers

JPMorgan Chase CEO Jamie Dimon has issued a sharp warning about “rogue” AI agents, arguing that the technology should not be allowed to speak for customers in the way a person or authorized system would. The comment, reported by Yahoo Finance, comes as banks and other large institutions race to apply artificial intelligence to customer service, operations, and fraud detection, even while regulators and executives question how far automation should go.

Dimon’s concern, as characterized in the report, centers on the risk that an AI system could act without appropriate permission or fail in ways that are hard to detect in real time. In banking, phone calls and messaging can trigger high-stakes actions, from account changes to dispute handling, and a “call on behalf of customers” implies more than answering questions. It suggests decision-making and execution under a customer’s identity.

While the report focuses on the warning itself, it does not lay out specific internal policies JPMorgan may be applying, nor does it provide details on what kinds of AI agents JPMorgan is using today. The absence of operational specifics matters, because Dimon’s remarks could be interpreted as a general stance against over-automation, or as a targeted concern about particular workflows where authorization and intent are unclear.

The broader issue in finance is that AI agents are often described as capable of taking actions, not just generating responses. An “agent” can be built to interpret requests, decide on a next step, and interact with systems that can move money, update records, or escalate cases. For executives, the practical question is whether the agent can demonstrate that it has the right to act, and whether its actions can be audited and corrected quickly if something goes wrong.

Regulators have increasingly emphasized governance for AI in financial services, including controls around model risk, data privacy, and oversight of automated decisions. Even without naming a regulator or a rule in the report, Dimon’s comments fit into a wider industry debate about accountability when an AI system behaves unexpectedly, produces incorrect information, or follows an instruction that conflicts with a customer’s intent.

For JPMorgan, the caution also reflects a brand and risk-management reality. A bank’s customer communications are tightly managed, both to protect customers and to reduce operational and legal exposure. Letting an AI agent initiate calls or represent customers could increase the number of touchpoints where verification fails, or where a customer misunderstands what they are authorizing.

It is also unclear from the reported remarks whether Dimon is opposed to all AI assistance, or specifically to AI agents that can contact people or respond to inbound requests without human confirmation. The distinction matters because many banks have already deployed AI tools that summarize accounts, draft messages, or route inquiries, while keeping final authorization with people or well-scoped systems.

What to watch next is whether JPMorgan clarifies its AI-agent governance approach, including how it handles authorization, escalation, and monitoring for customer communications. Investors and customers will be watching for any concrete policy statements about when AI may speak for a customer, what “rogue” behavior JPMorgan is trying to prevent, and how failures would be detected and corrected. If more detail emerges, it could become a announcement for how cautious the largest U.S. banks intend to be as AI agents move from assistance to action.