Who's liable when an AI agent uses your login?
Who is liable when an AI agent uses your bank login? Wells Fargo and Chase now say the customer is. What the terms say and the read-only way to stay clear.
If an AI agent uses your bank login, the banks now say you are liable for what it does. Wells Fargo's new terms, from 24 November 2026, treat an agent's instructions as yours, and Chase's updated agreement, from 1 January 2027, holds customers responsible for an agent's actions. The way to stay clear is never to give an agent your login.
Schwab's chief executive Rick Wurster asked the question first, about Meta's Muse: "We don't want someone just giving their password to Muse… who's liable?" Banks have now answered it in writing, and the answer is the person who shared the password.
Who is liable if an AI agent uses my bank login?
You are, under the terms Wells Fargo and Chase have written. Wells Fargo treats an AI agent's instructions as yours, and Chase's agreement holds customers responsible for what an agent does with their access. Bank of America's customer email says the same in a sentence: share your credentials and "you're responsible for activity resulting from that access."
The reasoning is the same one that has always applied to sharing a password. The bank sees your login, so it treats the activity as yours. What is new is that the terms now name AI agents, which closes the gap a customer might have hoped to find, and makes the responsibility explicit before anything goes wrong.
What do Wells Fargo's terms say?
Wells Fargo's Online Access Agreement takes effect on 24 November 2026. It treats instructions from an AI agent as the customer's own, says the customer is responsible for "properly and timely revoking any authorization provided to an AI Agent", and allows the bank to block AI tools.
There are three duties in it: own what the agent does, revoke it, and expect blocking.
The revoking duty is the one to notice. It makes you the record-keeper of every agent you have ever authorised, and you answer for any you forgot. Wells Fargo's AI agent terms, explained goes through the wording clause by clause.
What do Chase's terms say?
Chase's updated Digital Services Agreement takes effect on 1 January 2027 and holds customers responsible for an agent's actions. Chase has also said it does not permit third parties to access accounts by obtaining customers' banking credentials, while read-only access through open-banking providers is allowed.
So the same bank draws a clear line between a password and a permitted read-only link.
Chase's AI agent terms: what changes in 2027 sets out what you would want to do before the date.
What about Citi, Bank of America and Schwab?
Citi says it does not permit third-party AI agents or assistants to log in to a customer's account or transact directly within its digital banking platforms. Bank of America puts the responsibility on the customer who shares credentials. Schwab blocks unauthorized agents and has promised its own MCP servers for clients who want to connect their assistants.
None describes a safe way to hand an agent a password.
The full list of banks, with what each said and when, is in which banks block AI agents.
How do I keep from being liable?
Keep your password out of every assistant and use a read-only connection that you approve in your bank's own sign-in. A connection of that kind can't act as you, so there is no agent activity in your name to be liable for, and you can end it whenever you like.
If you have shared a password with an assistant already, change it.
Era is built around that. You link your bank once through a regulated data network, Era holds read access that you can revoke, and every assistant connects to Era. This is why one hub is easier to manage than a link per assistant: there is one list of what has access, and one place to switch it off. You can:
- Keep keys read-only, with an expiry date you choose.
- See which assistants are connected and revoke one without touching the rest.
- Disconnect a bank to end access for every assistant at once.
- Check an edit trail of what any assistant has changed.
How to audit an AI agent managing your money walks through those controls, and is it safe to connect your bank to ChatGPT or Claude covers the wider safety questions.
What should I do before the dates arrive?
Make a list of every assistant, app or service you have ever given a banking login, and replace each with a read-only link. Do it before 24 November if you bank with Wells Fargo, and before 1 January 2027 if you use Chase. Then link your accounts once in Era and connect your assistants to Era.
Era connects accounts across the US, the UK, Canada, France, Germany, Spain, and 40+ countries in all. Moving money between your own accounts is in private beta, off by default for every key, and every transfer needs your approval.
Related reading
Last updated