JPMorgan CEO Jamie Dimon Warns AI Agents Could Make Unauthorized Financial Decisions as Bank Blocks Outside Bots

Context

As technology companies including Meta roll out autonomous consumer AI assistants such as Muse with financial integrations, banks face growing pressure over how third party agents interact with customer accounts. JPMorgan Chase, the largest US bank with $5.1 trillion in assets, invests roughly $2 billion annually in internal AI systems used by 150,000 employees each week, while piloting its own internal cash management tool called Smart Cash.

Deep dive

Polymarket relayed remarks from JPMorgan Chase CEO Jamie Dimon warning that autonomous AI agents given access to bank accounts could make unauthorized or improper financial decisions. Note that Polymarket's post condenses remarks Dimon delivered at the Economic Club of Chicago alongside comments from JPMorgan's third quarter earnings call, as reported by Business Insider and CNBC.

Speaking at the Economic Club of Chicago, Dimon compared the speed of AI development to a tsunami and said that even governments cannot halt its global spread. Addressing autonomous agents given access to bank accounts to make payments on a user's behalf, Dimon warned that once agents are inside a customer's bank account, users are assuming the agents will act properly when they may not. He added regarding the rapid spread and cybersecurity risks of the technology, "You can't stop it. You just have to, like, manage the hell out of it."

Those remarks align with JPMorgan's operational policy confirmed to CNBC and Fortune. On the bank's third quarter earnings call, Dimon stated that JPMorgan will not allow outside agents to log into customer accounts or execute actions on their behalf, warning of the risk of thousands of rogue agents operating at scale and noting that customers who hand credentials to third party tools face a liability transfer. A JPMorgan spokesperson confirmed that while customer authorized open banking connectors such as Plaid can read balances in a read only format, third party AI agents are blocked from moving funds or paying bills.

What to expect next

Financial institutions, fintech data aggregators, and AI labs will need to establish cryptographic agent authentication, scoped transaction permissions, and clear fraud liability standards before major retail banks allow external AI agents to move customer money autonomously. Until regulators and payment networks formalize rules for delegated agentic commerce, large banks are likely to favor proprietary internal tools like JPMorgan's Smart Cash pilot over third party consumer bots.