Bank of America Warns Customers Against Sharing Passwords With AI Assistants
Table of Contents
You might want to know
- Why does Bank of America caution customers against giving their login details to AI agents?
- Who may be responsible if an AI assistant uses shared banking credentials to access an account?
Main Topic
Bank of America is reminding customers to protect their financial accounts by keeping their banking credentials private, including from AI assistants. In an email to customers titled “Protecting your finances in the AI world,” the bank advised people never to share their usernames and passwords with another person or an outside service. Its warning explicitly includes “an AI-enabled tool or agent that offers to act on your behalf.”
The message reflects a growing tension between the convenience promised by personal AI assistants and the security requirements of financial services. Some AI tools are designed to perform tasks for users, and accessing an account may seem like a straightforward way to let an assistant check balances, manage purchases, or complete transactions. But giving an agent a banking username and password also gives it a direct route to sensitive financial information and account functions.
One possible consequence is that an AI assistant could take an action involving a customer’s money without that customer realizing what has happened. Even if an agent is intended to follow instructions, it may interpret a request incorrectly, act in an unexpected way, or perform a task that the account holder did not mean to authorize. The concern is not limited to deliberate misuse: mistakes, unclear instructions, or unexpected interactions with other services could also create problems.
Sharing credentials may also trigger a bank’s security systems. If an account is accessed in a way that appears unusual, the bank may suspect that an unauthorized person is attempting to take control. It could then restrict access or lock the customer out while investigating. Such measures are meant to protect customers, but they can be disruptive when a legitimate account holder has handed credentials to an outside tool and the resulting activity looks suspicious.
Another risk is that the AI service itself could have a security flaw. If an assistant stores, processes, or transmits login information insecurely, a vulnerability could expose those credentials beyond the service. Someone with financial motives could then use the leaked information to try to access the customer’s bank account. Sharing a password with an AI tool can therefore create risks both through the agent’s actions and through the possibility that the credentials are exposed.
These concerns are emerging as banks and other financial companies explore how AI agents might operate in the financial system. Banks are still working out how customers could safely use AI tools to access accounts and carry out tasks. The broader question includes agentic payments, in which people might configure an agent to make purchases autonomously at the best price or at an appropriate time. This type of automation could be useful, but it also raises questions about authorization, oversight, responsibility, and the safeguards needed to prevent errors or misuse.
As discussed onstage at TechCrunch Disrupt, banks, payment providers, and processors are still trying to determine how to enable account access through AI tools. The challenge is not simply whether an agent can complete a task. Financial institutions also need to consider how to confirm that the customer has authorized it, how to limit what it can do, how to monitor its activity, and how to respond if something goes wrong. The warning from Bank of America underscores that these arrangements have not yet removed the risks associated with sharing a password.
Bank of America did not immediately respond to a request for comment about the email. Its message nevertheless gives customers a clear instruction: do not share banking credentials with people, outside services, or AI-enabled agents that offer to act on your behalf. The bank also highlighted the potential consequences of choosing to share them. “If you do share your credentials,” the email reads, “keep in mind that you’re responsible for activity resulting from that access.”
That statement matters because account agreements at most banks make customers responsible for people or services to whom they provide their login information. Customers who hand over credentials may therefore face more than the practical risk of an unwanted transaction or a temporary account lock. They may also have difficulty disputing activity that results from access they chose to grant. The bank’s warning places responsibility for credential sharing squarely on the customer.
The safest approach, based on the warning, is to keep bank usernames and passwords out of AI assistants and other outside services. Customers considering automated financial tools should pay close attention to how the service obtains permission, what access it receives, and who is accountable for activity conducted through it. Until banks and technology providers establish secure, clearly authorized ways for agents to interact with financial accounts, handing over a password may expose users to risks that are difficult to predict or control.
The email was reported via Bryan Toporek. Its message arrives at a time when interest in AI-powered assistance is expanding, while the rules and protections for AI agents in financial services are still taking shape. For customers, convenience does not eliminate the need to protect credentials or understand the consequences of granting account access.
Key Insights Table
| Aspect | Description |
|---|---|
| Bank’s guidance | Do not share banking usernames or passwords with other people, outside services, or AI-enabled tools and agents. |
| Unauthorized activity | An AI assistant could take financial actions without the customer’s knowledge or interpret instructions in an unintended way. |
| Account access | Unusual activity could lead a bank to suspect an account is under attack and restrict or lock access. |
| Security exposure | A flaw in an AI service could expose credentials and potentially give financially motivated hackers a route to an account. |
| Customer responsibility | Bank of America’s email says customers are responsible for activity resulting from access granted by sharing credentials. |
| Industry challenge | Banks, payment providers, and processors are still determining how to enable secure AI-assisted account access and agentic payments. |
Afterwards...
AI agents may eventually become more capable of managing purchases and other financial tasks, but their use will depend on clear authorization, limited access, reliable safeguards, and transparent responsibility when things go wrong. Banks and technology providers still have work to do before customers can confidently delegate financial activity. In the meantime, Bank of America’s practical advice is straightforward: keep banking credentials private, and do not assume that an AI assistant can safely handle them on your behalf.
Last edited at：2026/10/28
