Bank of America is begging customers not to hand their passwords to AI assistants

AI assistants can write your emails, book appointments, shop on your behalf, and increasingly handle tasks you’d rather not do yourself. It’s the kind of convenience that’s hard to resist, even as questions grow about how much access we’re giving these tools. However, letting an AI assistant manage your calendar is one thing. Trusting it with your life savings is quite another, and Bank of America would rather its customers didn’t find out the hard way.

The banking giant has emailed customers asking them not to share their banking credentials with AI assistants or agents. It’s a curious warning from a company that, in the same message, celebrates its investments in AI to make banking easier, more personalized, and more secure.

The email, aptly titled “Protecting your finances in the AI World,” was shared on Bluesky by Bryan Toporek. It specifically cautions against sharing usernames and passwords with outside services, including AI agents that offer to act on customers’ behalf. A Bank of America spokesperson confirmed the email’s authenticity to TechCrunch.

Then comes the part that might make you reconsider letting your favorite chatbot anywhere near your savings: “If you do share your credentials, keep in mind that you’re responsible for activity resulting from that access.”

In other words, your AI assistant might do the clicking, but you could be left dealing with the consequences. Though that warning doesn’t override federal protections for genuinely unauthorized electronic transfers.

AI agents are getting smarter, and banks are getting nervous

The warning isn’t happening in a vacuum. Google is making Gemini a more capable assistant, while rivals like Claude Duet can now manage tasks across connected apps. OpenAI is in the race, too, with ChatGPT Work handling complex tasks across apps and files, and its newer Dots designed to keep working even when you’re away.

The appeal is obvious: why juggle multiple apps when an assistant can handle the legwork? Unfortunately, your checking account isn’t the ideal playground for an AI experiment, and a few recent cases lend some credence to this email warning.

An Axios investigation recently revealed that major US banks have detected unapproved third-party agents accessing customer accounts, sometimes acting beyond users’ instructions. Morgan Stanley has reportedly begun blocking such agents, while JPMorganChase and Citi are also restricting access. Chase has similarly discouraged customers from sharing their login credentials.

The problem isn’t necessarily AI itself. Banks struggle to distinguish between an agent following legitimate instructions and one acting without proper authorization. And when things go wrong, there’s the thornier question of who takes responsibility: the customer who granted access, the company behind the AI agent, or the bank that allowed the transaction? It’s a liability headache the financial industry still hasn’t fully figured out.

There is a middle ground, though. AI can help analyze spending, build budgets, or explain financial documents without needing your banking password. Where supported, bank-authorized read-only connections offer another option without handing an assistant the ability to move money.

But for now, the safest arrangement might be to let AI help plan your spending while you keep control of the actual spending.