Jamie Dimon drew a hard line this week: outside AI agents, including Meta’s Muse, will not be allowed inside JPMorgan Chase customer accounts. Not to check a balance for you. Not to pay your bills. Not at all.
The biggest bank in America is done pretending the agentic AI era is someone else’s problem. And it isn’t alone.
“We’re not going to allow just any … outside agent to come in inside the company, because it’s just too risky for clients and for JPMorgan,” Dimon told analysts on the bank’s third-quarter earnings call Tuesday, CNBC reported. Then the number that explains the urgency: “You don’t want rogue agents, and it could be thousands of them.”
How the JPMorgan block works
The mechanics are narrower than a total ban, but they kill the use cases that matter. JPMorgan confirmed to CNBC that it blocks external AI agents from acting on behalf of retail customers: an agent can’t log in with your banking credentials and it can’t move money. What still works: read-only access, where an agent sees balances through authorized open-banking connectors like Plaid. Your password stays out of it.
“Because it is not safe, we do not permit third parties to access accounts by obtaining customers’ banking credentials and logging in on their behalf,” a JPMorgan spokeswoman said. She gave no timeline for when, or whether, direct account access might open up.
The bank even named the casualties. Per CNBC, the restriction covers assistants such as Meta’s Muse and Instinct, the agent startup from Spear Street Technology. Muse went viral after its September launch as a personal AI assistant that can act across apps; Amazon already blocked it from shopping on Amazon.com over unauthorized access and credential concerns. The banking industry is now reading from the same playbook.
Citi is drawing the same line
Citigroup is restricting AI agents from accessing customer accounts too, the Financial Times reported on October 16. Citi’s version has a different flavor than JPMorgan’s outright block, but the direction is identical.
The clearest window into Citi’s thinking came from CEO Jane Fraser at the Sibos 2026 banking conference in Miami on September 30. Fraser argued that autonomous agents need airtight identity verification and authentication before they can execute transactions, whether they’re acting for a customer or for the bank itself. “I am confident I won’t have wild agents without the right controls in place,” she said, according to Fund Society’s coverage of the keynote.
Behind the scenes, Citi runs an internal governance platform called ARC, introduced this year, that every internal AI agent must pass through. “You cannot create an agent without going through ARC,” the bank says. It’s the difference between Dimon’s wall and Fraser’s checkpoint system, but both keep outside agents away from real money.
The real question is liability
Dimon’s most quotable line wasn’t about rogue agents. It was aimed at customers: “If people are using them, they should understand the risk,” he said. “They should understand there’s a liability transfer.”
Read that twice. The bank is telling you, on an earnings call, that if you hand your credentials to an AI agent and it sends money somewhere you never intended, that’s on you. Banks spent the last two years selling investors on AI agents running the back office. Now the same executives are ring-fencing the one door they control — the customer’s account — and the legal logic is plain: when a human makes a mistake, the bank absorbs the fraud cost. When thousands of agents make the same bad call at once, that’s a systemic loss nobody wants to underwrite.
The business stakes go beyond fraud. AI agents that automatically sweep idle cash out of low-yield deposit accounts into higher-yielding alternatives would erode the cheap deposit base commercial banks depend on to issue loans. The Times of India reported that Apollo chief economist Torsten Slok puts roughly $1.6 trillion in non-interest-bearing deposits at JPMorgan Chase, Bank of America and Wells Fargo in the blast radius if agents routinely reallocate consumer cash. Banks are blocking the agents. It’s not hard to see both motives at work.
Who owns the customer when an agent sits in between
Here’s the part neither bank says out loud: the lockout is also a product strategy. JPMorgan isn’t anti-agent. Chief Financial Officer Jeremy Barnum used the same earnings call to tout Smart Cash, the bank’s own pilot AI tool that predicts a customer’s cash flow and moves funds between accounts to maximize yield. Agents, yes, but theirs.
Citi is doing the same dance. It told investors on October 13 that an AI coding tool now used by 14,000 of its engineers has cut tech vendor spending by 30% year over year. The message is consistent across Wall Street: build AI in-house, where the governance platform, the audit trail, and the liability all live under one roof.
For the startups building agentic personal-finance tools, this is a meaningful headwind. The most compelling pitch, an agent that can see and act on all your money, just lost access to the largest banks in the country. Watch for two things next: whether smaller banks follow the big names, and whether regulators step in to define what counts as legal authorization when software moves your money. Until then, if you want an AI to touch your checking account, it’ll have to be the bank’s own.
Frequently asked questions
Which banks are blocking AI agents from customer accounts? JPMorgan Chase blocks outside AI agents from acting on customer accounts, and Citigroup is restricting agent access too, according to the Financial Times. Bank of America is also tightening limits, and Amazon previously blocked Meta’s Muse from its site.
Can AI agents still see my bank balance? Yes. JPMorgan allows read-only data sharing through authorized open-banking connectors like Plaid. What’s blocked is agents logging in with your credentials or moving money and paying bills.
Why do banks block AI agents? Banks cite security, fraud risk, and liability. Jamie Dimon warns of “rogue agents” operating at scale and says customers who delegate to outside agents accept a liability transfer if something goes wrong.
Will banks ever let AI agents move money for me? JPMorgan’s spokeswoman said there is no timeline for allowing direct account access, and both banks are building their own AI tools instead. JPMorgan is piloting Smart Cash, which predicts cash flow and moves funds between accounts to maximize yield.
Sources: CNBC, Financial Times, Fund Society, The Times of India.