Major U.S. banks, including JPMorgan Chase and Morgan Stanley, are blocking or restricting outside AI agents from customer accounts.
I think that’s the right call for now. But customers will keep wanting agents to pay bills and manage their money. Banks will have to adapt.
Sharing a bank login gives an agent access that may go beyond what the customer intended. The bank has to work out who is behind the agent, what it was asked to do and when to stop it.
Know Your Agent (KYA) is important. Banks need to know who controls an agent, limit what it can do and be able to cut off its access.
I also think KYA will be a game of whack-a-mole for a long time. New agents and ways to misuse them will keep appearing. Banks will have to keep updating their checks as both customers and scammers find new uses for agents.
But even an agent with the right permissions can make a dangerous payment. That brings us to the question I’d put first:
What’s the worst thing an AI agent can do with your bank account?
Send your money to someone who should never get it.
Imagine asking your agent to pay your bills. It gets an email that looks like it came from your internet provider: “We’ve changed banks. Please use these new account details.” The email is fake. Your agent believes it and pays.
You gave the agent permission to pay your bills. Knowing who controls it and checking that permission wouldn’t have caught the fake invoice.
Banks already face this with people. A scammer convinces someone to send money from their own phone, using their own account. The bank recognizes the customer and sees them approve the transfer. Everything can look normal.
Asking the customer to approve it again may not help. If they still believe the scammer, they’ll say yes again. An agent can fall for a fake request too, or follow an instruction from a customer who is being tricked.
That’s why I’d put stopping scam payments ahead of identifying agents. Both matter, but permission alone doesn’t tell a bank whether the money is going to the right person.
Look at who’s getting the money
The bank needs to understand the account receiving the money: its connections to known scams and any signs that the sender and receiver know each other. Those clues can help whoever is making the payment.
That’s what we built Incognia’s Real-Time Scam Prevention to help banks do. Our Physical Social Graph uses the places devices have visited over time to help reveal those connections. Our indoor location technology is precise enough to tell apart neighboring apartments.
For example, several accounts might look unrelated to a bank. But if the phones using them keep showing up in the same apartment, that could point to one scam operation. Some may be “mule accounts,” used to collect and move stolen money.
The same history can help explain a payment that looks unusual. Two people may live far apart but regularly visit each other. Being in different cities doesn’t mean they’re strangers.
We found that nearly 79.2% of legitimate payments were between people with a real-world connection. By comparison, 86.2% of reported scams involved people with no such connection.
Of course, sharing a place doesn’t prove that someone is trustworthy. And people pay strangers for perfectly good reasons. We combine these clues with other warning signs, such as someone controlling the customer’s phone remotely or being on a call while making the payment.
In an analysis of transactions from a bank with more than 80 million active users, we identified approximately 70% of scam transactions.
The bank gets an assessment of the risk and the clues behind it. Alongside its own checks, it can use these to let a payment through, warn the customer or stop it. These checks depend on having the relevant device and location history.
To me, that’s the most important job of “intent intelligence”: helping a bank catch a scam before the money is gone, whether the payment comes from a customer, their agent or a fraudster’s agent.
Banks should keep improving agent controls. They can also use what they know about scams today to protect payments made by agents. If an agent is about to send your savings to a scammer, you want your bank to stop the payment.