If you hand your bank login to an AI agent, anything it does with your money is on you. Bank of America sent customers that message in a recent email titled Protecting your finances in the AI World, according to TechCrunch AI. The bank told customers never to share their username and password with another person or an outside service, “including an AI-enabled tool or agent that offers to act on your behalf.”
A Bank of America spokesperson confirmed the email is real after it circulated on social media. That makes this one of the clearest public warnings a major US bank has given about personal AI agents.
What the Bank Actually Said
The main point is about who’s liable. Most account agreements already make customers responsible for anyone they give their login to, and Bank of America says so plainly in the email.
“If you do share your credentials,” the email reads, “keep in mind that you’re responsible for activity resulting from that access.”
In plain terms, if your AI assistant makes a bad transfer, buys the wrong thing or gets hijacked, don’t count on the bank to make you whole. TechCrunch AI notes that “you can practically sense the frustration” in the message. It reads like a bank that has already seen this go wrong more than once.
Why Sharing Your Login Is Risky
Personal AI assistants keep getting more capable, and more people are giving them full access to their accounts. TechCrunch AI lists three ways this can go wrong:
- The agent acts without you knowing. An autonomous agent with your full credentials can move money or make purchases you never approved.
- The bank locks you out. A login from unfamiliar infrastructure can look exactly like an account takeover, so the bank’s fraud systems may freeze your account.
- Your credentials leak. If the AI tool has a security flaw, your banking login could end up with financially motivated hackers.
The third risk worries me most. A password you give an agent doesn’t just sit in your head anymore. It now lives in someone else’s system, and you can’t control how that system is secured, logged or attacked. You’re trusting every link in that chain.
Banks Are Stuck in the Middle
The warning comes as the financial industry struggles with something it doesn’t have a clean answer for yet. Banks are still working out where AI agents fit, and TechCrunch AI reports they’re dealing with unauthorized agents getting into customer accounts.
The pressure is real. Banks, payment providers and processors are trying to figure out how customers can safely connect AI tools to their accounts. One big use case is agentic payments, where you tell an agent to buy something automatically at the best price or the right moment. TechCrunch AI covered this tension onstage at its Disrupt conference.
Agents are moving faster than the rules. Customers want AI that can actually do things with their money, and many AI tools offer that by asking for a username and password, because that’s the easiest way in. Banks don’t control that kind of access, can’t easily audit it and never designed for it. Until there are proper permission-based connections that are widely available, credential sharing will fill the gap, and customers will carry the risk.
What You Should Do Now
If you’re experimenting with AI assistants for personal finance, play it safe:
- Don’t type your bank password into any AI tool. That includes browser agents and “do it for me” assistants.
- If you already have, change it today. Then turn on multi-factor authentication if it isn’t already on.
- Use official connections. If your bank offers an approved way to link third-party apps, use that instead of giving out your raw login.
- Turn on transaction alerts. Real-time notifications help you catch activity you didn’t approve before it adds up.
- Read your account agreement. Know where your bank draws the line on liability before something goes wrong.
If you’re building AI agents, take this as a warning too. Products that depend on users handing over their passwords are on shaky ground, and banks are now telling customers so directly.
What Comes Next
Expect more banks to send warnings like this, and expect pressure to grow for standard, permission-based ways to connect AI agents to financial accounts. Agentic payments are coming whether banks are ready or not. The real question is whether safe, approved access arrives before a high-profile agent breach forces the issue. Until then, keep your password to yourself. You can find more details in the original TechCrunch AI report.