Bank of America has warned its customers never to share their banking username and password with an AI assistant or agent, saying that handing over login credentials puts their money at risk even when the tool promises to act on their behalf, TechCrunch reported on October 27, 2026. The warning came in a customer email titled “Protecting your finances in the AI World,” whose authenticity a Bank of America spokesperson confirmed to TechCrunch after the email was posted on social media. The message is blunt: if you share your credentials, you are responsible for whatever activity follows from that access.
A Warning, Not A New Product
The email urges 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.” That phrasing is deliberate. It does not single out one chatbot or one app. It covers the whole emerging category of personal AI assistants that offer to check balances, pay bills, find deals or complete purchases for you — tasks that are only possible if the assistant can log in as you.
Customers are increasingly handing over those logins as personal AI assistants grow more popular, TechCrunch noted. The appeal is obvious: an assistant that can see your accounts can do more for you than one that cannot. The bank’s argument is that the convenience hides three distinct risks, and that customers, not the assistant, carry them. First, an assistant could do something with your finances without your knowledge. Second, the bank itself might interpret the assistant’s activity as someone raiding your account and lock you out. Third, a security flaw in the assistant could leak your credentials elsewhere, giving financially motivated hackers a way into your bank account.
Why The Liability Line Matters
“If you do share your credentials,” the bank’s email reads, “keep in mind that you’re responsible for activity resulting from that access.” Account agreements at most banks hold customers responsible for anyone they give their login to, TechCrunch added. That is the part of the warning most likely to surprise users. People tend to think of an AI assistant as a tool, like a browser, rather than as a person they have authorised. Banks, in contractual terms, treat it differently: once you hand over the keys, the actions taken with those keys are treated as yours.
That distinction is becoming harder to ignore as agentic AI spreads. Banks, payment providers and processors are still trying to figure out how to allow customers to access bank accounts using AI tools, including for agentic payments, where agents autonomously buy things at the best price or time, as discussed at TechCrunch Disrupt. The industry wants those use cases, but does not yet have a settled, secure way to grant an agent limited, revocable access without handing over full login credentials.
Agents Are Already Knocking On Bank Doors
Bank of America’s warning comes as banks are reportedly grappling with unauthorised agents accessing customers’ money, TechCrunch reported, citing Axios. The tension is visible across the sector. On one side, customers want assistants that can act across their financial lives. On the other, banks need to distinguish a legitimate customer instruction from an attacker, a buggy agent or a third party that has quietly harvested credentials. Password sharing collapses that distinction, because to the bank’s systems an agent logging in with your password looks exactly like you.
The same identity problem is driving work on verified agent identities, as in our report on Socure’s verified identity for AI agents and our report on IBM giving every AI agent an identity and a kill switch. Those efforts point to the long-term answer: agents with their own credentials, limited permissions and an off switch, rather than agents borrowing their owner’s password.
Why It Matters
This matters because the password-sharing habit, once formed, is hard to break — and it undermines every other protection a bank offers. Fraud detection, transaction alerts and account locks all assume that activity from your login is activity by you. If an assistant, or an attacker who compromised that assistant, uses the same login, those protections lose their signal. The risk is not theoretical: a leaked credential does not just expose one account, it exposes every service where that assistant was given access.
It also matters for the future of agentic payments. The promise of agents that shop, pay and manage money for you depends on trust, and trust depends on clear lines of responsibility when something goes wrong. Bank of America’s email draws that line in the bank’s favour for now, as in our report on unvetted AI agents slipping into US bank accounts. Until agents get their own verifiable identities and limited permissions, customers who want AI help with their finances face a choice: convenience with full credential sharing, or safety with the assistant kept outside the vault.
Background
Bank of America sent the warning as an email blast to customers titled “Protecting your finances in the AI World,” and a spokesperson confirmed its authenticity after it was posted on social media. The bank’s position restates a long-standing rule — never share banking credentials — now applied explicitly to AI-enabled tools and agents that offer to act on a customer’s behalf.
Source: TechCrunch report
0 Comments