🔍 Read the full analysis: 7 Ways To Let An AI Agent Handle Money More Safely on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Bank of America has warned customers not to share banking credentials with AI tools, saying customers are responsible for activity resulting from that access. Safer options include read-only permissions, payment approvals, capped virtual cards and separate accounts, but legal responsibility for agent actions is not fully settled.
Bank of America has warned customers not to give their banking passwords to AI agents, saying people who share credentials are responsible for activity resulting from that access. The warning comes as financial companies report growing use of agents for payments and trading, while questions about security and who bears losses remain unresolved.
The bank’s email, titled “Protecting your finances in the AI World,” says usernames and passwords should not be shared with another person or provided to an outside service, including an AI-enabled tool or agent. Bank of America said it is “excited about what AI used responsibly can deliver.” A customer posted the email on Bluesky; TechCrunch reported it on October 27, and a bank spokesperson confirmed its authenticity to the publication.
The warning is landing amid mixed signals from the financial industry. Visa told investors this week that trust is the biggest barrier to AI agents paying for goods. Robinhood said customers have placed more than $250 million into accounts that AI agents trade. Those figures indicate activity is already taking place, but they do not establish how many customers use agents or how often agents make transactions.
The source article identifies several safer alternatives to sharing a master password: read-only access, requiring approval for every payment, using a virtual card with a spending cap, and giving an agent a narrowly scoped token limited by merchant, amount and time. A separate account or pot of money can also limit potential losses. These controls reduce access; they do not by themselves settle who is liable if an agent acts improperly.
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
Limits Can Reduce Agent Payment Risk
A banking password can let software act with the same broad access as the customer. That creates two distinct risks: an agent may make an unintended transaction, and a bank may be unable to distinguish an authorized agent from someone who has stolen the customer’s credentials. The source article says banks have been blocking third-party agents and revising terms as they confront account activity they cannot readily classify.
There is also a potential cost to customers. The source article discusses US and EU payment rules, but stresses that applying those rules to an AI agent has not been tested in court, to the author’s knowledge. Technical safeguards are not a guarantee of reimbursement, and customers should not assume that a bank or brokerage will cover every loss caused by an agent they authorized.
For readers, the practical distinction is between letting an agent perform a narrow task and giving it account-wide power. Approval steps, low limits and separate funds can constrain what an agent can do if it misunderstands an instruction or is manipulated. The level of protection depends on how the specific service works and what its terms say.
Products Worth Considering
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Trezor Safe 5 Crypto Hardware Wallet with Color Touchscreen, Black Graphite
Trezor Safe 7 Crypto Hardware Wallet
As an affiliate, we earn on qualifying purchases.
Banks Face Growing Agent Access
The warning follows a report by Axios on October 22, citing unnamed sources, that major US banks were finding third-party AI agents inside customer accounts, often using customers’ own passwords. According to that report, some banks could not determine whether the activity reflected what customers wanted. The report is attributed to unnamed sources, and the source material does not provide a public count of affected accounts.
One technical risk is prompt injection: instructions embedded in a webpage, email or file may steer an AI agent that reads that content. The source article describes prompt injection as the top risk on OWASP’s list for AI applications. It does not establish that any particular bank transaction was caused by such an attack; it explains why broad access can be dangerous when an agent processes material from outside sources.
The article also summarizes relevant legal provisions while marking the limits of that analysis. In the United States, Regulation E excludes certain transfers initiated by a person given account access by the consumer. In the European Union, the Payment Services Directive requires customers to take reasonable steps to protect personalized security credentials. How those provisions apply to AI agents remains unsettled, according to the source author; the legal discussion is not a court ruling or a substitute for legal advice.
““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 regulator or court ruling applying the cited US rule to an AI agent is identified in the source material. It also says it has not found a court test of whether giving credentials to an agent would count as gross negligence under the EU rules. Those points are the author’s assessment of the available material, not settled legal conclusions.
Details about the reported bank activity remain limited. The source does not name the banks involved in the Axios report, quantify affected accounts, or describe particular losses. It is also unclear how widely customers use agent-specific controls such as scoped tokens, how different services define approval, and whether those protections change the outcome of a disputed transaction.
Check Access Before Connecting
Before connecting an AI agent to a financial account, customers can review the service’s permissions and terms, look for read-only modes, transaction approvals and hard spending limits, and consider keeping agent-controlled funds separate from everyday money. They should also check how to revoke access and contact their financial provider if credentials have already been shared.
Financial firms’ policies, product controls and legal interpretations may develop as agent use expands. The source material gives no announced timetable for new rules or specific product changes. For now, the key question is practical: what can the agent access, and who bears the loss if it acts incorrectly?
Key Questions
What did Bank of America tell customers?
The bank said customers should not share banking usernames or passwords with outside services, including AI agents, and said they are responsible for activity resulting from shared credentials.
What are safer ways to give an AI agent financial access?
Options described in the source include read-only permissions, approval for each payment, capped virtual cards, narrowly scoped tokens and separate funds with limited balances. Availability depends on the provider.
Does sharing a password mean a customer must cover every loss?
That cannot be concluded from the source material. It describes relevant US and EU rules but says their application to AI agents has not been tested in court, to the author’s knowledge. Provider terms and the facts of a case may matter.
What is prompt injection?
Prompt injection is a technique in which instructions placed in content an AI agent reads—such as a webpage, email or file—may steer its behavior. The source identifies it as a risk, not as the confirmed cause of a specific financial loss.
Source: ThorstenMeyerAI.com
