🔍 Read the full analysis: Make Money Tasks Safer When You Use An AI Agent on ThorstenMeyerAI.com
Get smart everyday buys 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 warned customers that sharing banking credentials with an AI-enabled tool or agent could leave them responsible for activity conducted through that access. The warning comes as financial firms report growing use of agents around payments and trading. Read-only permissions, per-payment approval and capped accounts can limit exposure, but do not automatically settle who is liable for an agent’s actions.
Bank of America has warned customers not to share their usernames and passwords with AI-enabled tools or agents, saying customers who do so are responsible for activity resulting from that access. The warning, confirmed by the bank to TechCrunch after a customer posted the email, comes as financial firms report growing use of AI agents in payment and trading activity, raising questions about account security and who bears losses if an agent makes an unauthorized transaction.
The email, titled “Protecting your finances in the AI World,” says 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 adds that customers who share credentials are “responsible for activity resulting from that access.” Bank of America also said it is “excited about what AI used responsibly can deliver.” The bank’s warning is advice to customers; it does not, by itself, determine how a particular dispute or transaction would be treated under the law.
The warning follows reports of agents accessing financial accounts using customers’ own login details. Axios reported on October 22, citing unnamed sources, that major U.S. banks were finding third-party AI agents inside customer accounts and that banks often could not tell whether the activity had been authorized by the account holder. TechCrunch reported the Bank of America email on October 27 and said a bank spokesperson confirmed it was authentic.
On the same Tuesday, Visa told investors that trust was the main barrier to AI agents making payments, according to the source material. Robinhood said customers had put more than $250 million into accounts that AI agents trade. That figure is a reported amount in accounts, not a measure of trading profits or losses. Robinhood has also described using a separate account and spending limits, while saying customers “assume all risk” for trades made by AI agents.
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 Agent Access Changes Account Risk
A bank password is a broad credential, not a permission limited to one task. An agent that signs in as a customer may be able to see or do much of what the customer can do, while the bank may have difficulty distinguishing the agent from the customer or from someone who stole the password. That can create two kinds of problems: an agent could take an action the customer did not intend, or a bank could flag the login as suspicious and restrict access.
There is also a gap between technical safeguards and financial responsibility. A spending limit or separate account can restrict how much an agent can reach, but those measures do not necessarily establish who must cover a disputed transaction. Customers should review the provider’s terms and the bank’s rules before enabling access, particularly before allowing an agent to make payments or trades without approval.
Safer designs give an agent only the authority it needs. Options described in the source material include read-only access, approval for each payment, a virtual card with a fixed limit, or a token restricted to a specific merchant, amount and time period. A separate account with a limited balance can also contain potential losses. These are risk-reduction measures, not guarantees that an agent will behave as intended or that a provider will accept liability.
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
Brother TD-2020A 2‑Inch Direct Thermal Desktop Label Printer with USB
HP Sprocket Photo Booth Machine Instant Color Photo Printer with 10.1 Touchscreen, LED Ring Light, 250+ Frames, 3×4 ZINK Paper, QR Code Shar
As an affiliate, we earn on qualifying purchases.
Banks Face Unclear Agent Logins
AI agents can read and act on information from websites, emails and files. That creates a security concern known as prompt injection: instructions concealed in material an agent processes may influence what it does. The source article describes prompt injection as the top risk on the OWASP list for AI applications. The practical concern for money tasks is that an agent with broad account access may encounter instructions or content that steer it away from the user’s intended task.
The legal protections cited in the source material are not a definitive answer for every agent-related loss. In the United States, Regulation E excludes from its definition of an unauthorized electronic fund transfer one initiated by a person given access to the account by the consumer, subject to the regulation’s terms. Whether that provision applies to an AI agent or the company operating it has not been decided by a court or regulator, according to the source article.
In the European Union, the Payment Services Directive requires users to take reasonable steps to keep personalized security credentials safe. The source article says whether giving credentials to an AI agent would amount to gross negligence under the directive has not been tested in court. Those legal questions are distinct from the bank’s warning and should not be treated as settled conclusions about a customer’s rights.
“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 the customer email confirmed as authentic to TechCrunch
Liability Rules Remain Untested
It is not yet clear how courts or regulators would apply existing payment rules when a customer has intentionally supplied credentials to an AI agent, but the agent then makes a payment or trade the customer did not intend. The legal analysis in the source article is the author’s interpretation, not a ruling or legal advice. The reported bank activity also comes from unnamed sources cited by Axios, and the article does not identify the banks or provide counts of affected accounts.
The source material does not specify the exact limits, terms or customer protections attached to the Robinhood accounts behind the reported $250 million figure. Nor does it establish whether Visa’s comments reflect a particular product launch or a change in payment policy. A secure technical setup may reduce access, but the protections and responsibilities will depend on the provider, bank, transaction type and applicable rules.
Check Permissions Before Connecting
Before connecting an agent to a financial account, customers can check whether the service supports read-only access, per-transaction approval, merchant and amount limits, or a separate account with a restricted balance. They should also read the provider’s terms to understand how it assigns responsibility for agent-initiated activity and confirm with their financial institution what access methods it permits.
For companies building agent payment tools, the safeguards described in the source material point toward narrower permissions and clear approval steps instead of collecting a customer’s main banking password. Banks and payment companies will also need to explain how they identify authorized agent activity and how customers can report unwanted actions. The source material does not give a timetable for new rules or standards; for now, customers should treat broad account access as a significant exposure and keep approval controls in place.
Key Questions
What did Bank of America tell customers?
The bank said customers should not share their username and password with another person or an outside service, including an AI-enabled tool or agent. It said customers who do share credentials are responsible for activity resulting from that access.
Does giving an AI agent a password remove fraud protection?
That is not settled for every situation. The source article discusses U.S. and EU rules that may affect responsibility, but says the application to AI agents has not been tested in court or resolved by a regulator. A customer’s rights may depend on the facts, applicable law and account terms.
How can customers limit an agent’s access?
Possible controls include read-only permissions, approving each payment, using a capped virtual card, restricting a token by merchant, amount and time, or keeping the agent’s funds in a separate account with a limited balance. These measures can reduce exposure but do not guarantee against loss.
What is prompt injection?
Prompt injection is a security risk in which instructions in content an AI agent reads, such as a webpage, email or file, may steer its behavior. It matters for financial tasks because an agent with broad permissions could be influenced while handling account information or payments.
Source: ThorstenMeyerAI.com
