America's biggest banks want artificial intelligence to move money, serve customers and make operations more efficient. But as AI agents gain the ability to act on users' behalf, JPMorgan Chase & Co. (JPM) CEO Jamie Dimon is drawing a line.
"There's a defensive side, which is you've got to make sure our customers aren't hurt by it," Dimon said on JPMorgan's third-quarter earnings call Tuesday. He warned that autonomous tools could create a new category of risk.
"You don't want rogue agents, and there could be thousands of them," he said.
The concern is not simply that AI could make mistakes. Agents that interact with financial accounts may access sensitive information or initiate transactions. This raises questions about security, customer consent and who bears responsibility when something goes wrong.
Dimon wants customers who use these tools to be aware of risks, including "liability transfers" if their data is misused.
The bank is not rejecting the technology outright. CFO Jeremy Barnum said JPMorgan's Smart Cash solution, which helps customers manage cash across accounts, had entered an early pilot phase. The distinction is between deploying AI for a controlled customer use case and allowing outside agents broad access to banking infrastructure.
Citigroup Treats AI as a Cyber Risk
Citigroup, Inc. (C) is also strengthening cyber defenses.
"AI is tremendously exciting, but it also puts powerful tools in the hands of bad actors," CEO Jane Fraser said on Citi's earnings call. "If we're going to innovate at speed, we have to manage risk at speed and retain trust," she added.
Citi said it is testing its systems against disruptive scenarios and deploying advanced AI models within a controlled security framework. Fraser also emphasized coordination with peers and industry leaders, reflecting how interconnected financial systems can make security failures extend beyond a single institution.
For investors, that creates a second cost of AI adoption: banks must invest not only in tools that improve productivity and customer service, but also in safeguards that keep those tools from becoming a source of vulnerability.
Bank of America Expands Its AI Toolkit
Bank of America Corp (BAC), with CEO Brian Moynihan at the helm, reported more than 50 million active digital users, giving it a substantial customer base across which new capabilities could be deployed.
The three banks are approaching different parts of the same problem. JPMorgan is restricting access by external agents, Citi is emphasizing cyber resilience, and BofA is integrating AI into its broader digital-banking strategy.
For investors, the next test is whether banks can turn AI into measurable efficiency and better customer experiences while maintaining the security and trust that underpin their businesses. In banking, the technology's value may depend as much on what it is prevented from doing as on what it can accomplish.
