🔍 Read the full analysis: A Safer Setup For AI Agents That Work With Your Finances on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
Bank of America warned customers not to share their banking usernames or passwords with AI tools, saying customers are responsible for activity resulting from shared access. The warning comes as financial firms report growing use of agents for payments and trading, but the legal and practical risks of giving an agent account-level access remain unsettled.
Bank of America warned customers not to give their banking usernames or passwords to AI agents, saying customers are responsible for activity resulting from shared access. The warning arrives as financial firms discuss trust and report customers using AI agents to handle money, while questions about account security and who bears losses remain unresolved.
The email, titled “Protecting your finances in the AI World,” was posted by a customer on Bluesky and reported by TechCrunch on October 27. A Bank of America spokesperson confirmed to TechCrunch that the email was authentic. The bank said credentials “should never be shared with another person or provided to an outside service, including an AI-enabled tool or agent that offers to act on your behalf.” It added that customers who share credentials are responsible for activity resulting from that access, while saying it is “excited about what AI used responsibly can deliver.”
The warning comes amid reports of financial activity involving AI agents. Axios reported on October 22, citing unnamed sources, that major U.S. banks were finding third-party AI agents inside customer accounts, typically signed in with the customer’s own password. The report said banks may have difficulty distinguishing activity the customer authorized from activity they did not want. Separately, Visa told investors this week that trust is a major barrier to AI agents making purchases, according to the source article. Robinhood said customers had placed more than $250 million into accounts that AI agents trade; the source does not specify the measurement period or comparison baseline for that figure.
Giving an agent a password can grant it broad access under the customer’s identity, rather than permission for one limited task. That can leave a bank unable to tell whether a transaction came from its customer, a customer-directed agent or someone misusing the credentials. An agent may also be steered by instructions it encounters in web pages, emails or files, a risk known as prompt injection. The source article identifies prompt injection as the top risk on the OWASP list for AI applications.
Before your agent touches money
Give it a valet key, not the master key
Logging in with your password makes the agent you. A credential made for the agent can do only the job you gave it, and you can take it back.
- What it opensEverything you can do: transfers, payees, settings, statements
- LimitsNone that the agent cannot get around by logging in as you
- What the bank seesYou, or a thief. It cannot tell which
- If a web page tricks itThe trick reaches your whole account
- To take it backChange your password everywhere it was used
- If money goes missingUS rules can exclude transfers by someone you gave access to; Bank of America: “you’re responsible”
- What it opensOnly what you allowed: read-only, one card, one pot of money
- LimitsBuilt in: amount, merchant, time window, approval on your phone
- What the bank seesAn identified agent acting for an identified person
- If a web page tricks itThe damage stops at the limit
- To take it backDisconnect the agent or delete its card. Your password never moved
- If money goes missingDepends on the terms. Read them first: some still put all risk on you
A comparison of two set-ups, not a measurement. Which limits exist depends on your bank, card and app.
Sources: Bank of America customer email, confirmed to TechCrunch (27 Oct 2026); 12 CFR 1005.2(m); OWASP LLM01. Comparison: Thorsten Meyer.
Powered by Thorsten Meyer AI
Why Account-Level Access Raises Stakes
The immediate concern is not simply whether an agent can complete a payment or trade. It is what authority the agent receives, whether the customer can limit that authority, and who may bear the cost if the agent takes an unwanted action. A shared password can give software access that is difficult to distinguish from the account holder’s own activity.
That uncertainty matters for both consumers and banks. A customer may face an unauthorized transaction or an account lock if a bank treats unusual agent activity as a possible account takeover. The source article also notes that account protections may depend on whether a transfer is legally considered unauthorized, an issue that has not been settled for AI agents. A company’s security controls, such as spending limits, do not by themselves answer who is liable when something goes wrong.
Less-permissive access offers a way to narrow the risk: read-only permissions for tasks that do not require spending, approval before each payment, or a virtual card with a set limit. A scoped token could restrict an agent to a particular merchant, amount and time window. These approaches limit what an agent can do, though the source does not establish that any one method guarantees reimbursement or prevents misuse.
Popular right now
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Vitamix Propel Series 750 Professional-Grade Blender, 4 Programs, Black
Ninja SP151 Flip Toaster Oven and Air Fryer, 8-in-1, Stainless Steel
As an affiliate, we earn on qualifying purchases.
From Bank Logins to Limited Permissions
The bank’s message reflects a tension between the convenience promised by AI agents and the access they may need to act. A conventional login usually identifies the account holder, not the software operating on the holder’s behalf. When an agent uses a customer’s password, the institution may see what looks like a normal customer session even if the customer did not approve a particular transaction.
The source article describes U.S. and European rules that may be relevant, while stressing that applying them to AI agents is not settled. In the United States, Regulation E excludes some transfers from its definition of unauthorized when a person was given access to the account by the consumer. The author’s view that this could apply to an AI agent or the company operating it is an interpretation; the article says no regulator or court has ruled on that question as far as the author could determine.
In the European Union, the Payment Services Directive requires users to take reasonable steps to keep personalized security credentials safe. The source says it is unclear whether giving a password to an AI agent would count as gross negligence under the directive’s liability rules. That question, too, has not been tested in court according to the article. These legal points are not a determination of how an individual claim would be handled.
“Your username and password “should never be shared with another person or provided to an outside service, including an AI-enabled tool or agent that offers to act on your behalf.””
— Bank of America, in its customer email
Liability Rules Remain Untested
No court or regulator ruling on AI-agent credential sharing is identified in the source material. It is therefore unclear how consumer-protection rules would apply if an agent used a customer’s password to make a transaction the customer did not intend. The relevant facts may also differ between an agent acting within instructions and one that is misled or compromised.
The source reports Robinhood’s use of a separate account and spending limits, but says the company tells customers they “assume all risk” for trades by AI agents. The material does not provide the full terms, explain how those limits work, or say what happens in a dispute. It also does not establish whether every bank can detect or block agent activity, or whether the reported activity is widespread.
The $250 million figure attributed to Robinhood is reported without a time window or comparison baseline. Visa’s statement about trust is presented in the source as a comment to investors, but no fuller transcript or exact wording is provided. Readers should treat these as reported developments, not evidence that AI-agent finance is already safe or that losses are routinely covered.
Check Permissions Before Connecting
For customers, the practical next step is to read an agent’s access request and the financial provider’s terms before connecting an account. Prefer the narrowest permission that will do the job: read-only access where possible, explicit approval for each transfer, or spending restrictions and a separate account for agent activity. Customers should also check how to revoke access and how to report a transaction they did not authorize.
For companies building or offering financial agents, the source’s examples point to controls that limit permissions by task, merchant, amount and duration, rather than relying on a customer’s full login. Companies also need to explain clearly who is responsible for disputed transactions and what protections apply. A spending cap can reduce exposure, but it should not be presented as a substitute for clear liability terms.
The next developments to watch are how banks update their policies, whether regulators or courts address the treatment of agent-initiated transactions, and what specific safeguards financial platforms disclose. Until those questions are clearer, Bank of America’s warning is a concrete reminder: do not hand an AI agent a master key when a limited permission may be enough.
Key Questions
What did Bank of America tell customers?
It warned customers not to share their banking username or password with another person or an outside service, including an AI tool or agent. The bank said customers are responsible for activity resulting from credentials they share.
Does sharing a password mean a customer loses fraud protection?
That is not established by the source. It describes U.S. and EU rules that may affect liability, but says how those rules apply to AI agents has not been decided by a court or regulator. Customers should check their provider’s terms and seek qualified advice for a specific dispute.
What is a safer way to let an AI agent handle money?
Use limited permissions rather than a full account password where available. Options described in the source include read-only access, approval for every payment, a capped virtual card, a restricted token for a merchant and amount, or a separate account with limited funds.
What is prompt injection?
Prompt injection is an attempt to steer an AI system through instructions embedded in material it reads, such as a web page, email or file. The source identifies it as a risk for agents that process outside content, but does not claim that every agent is vulnerable in the same way.
What remains unknown about AI agents and financial accounts?
It is unclear how existing liability rules will apply to agent-driven transactions, how banks will distinguish customer-approved activity from misuse, and what protections platforms will provide when an agent makes an unwanted trade or payment.
Source: ThorstenMeyerAI.com
