🔍 Read the full analysis: Keep Your Bank Credentials Private When Using An AI Agent on ThorstenMeyerAI.com
Get the latest gadgets 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 their banking usernames or passwords with AI agents, saying they are responsible for activity that follows from shared access. The warning comes as financial firms report growing interaction between AI agents and customer accounts, while the legal treatment of agent-driven transactions remains unsettled.
Bank of America has warned customers not to share their banking usernames or passwords with an AI assistant, saying customers who do so are responsible for activity resulting from that access. The bank confirmed the email after a customer posted it online and TechCrunch reported on it on October 27, as financial companies confront growing use of AI agents to access accounts and make transactions.
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 acting on a customer’s behalf. It adds that customers who share credentials should keep in mind they are responsible for activity resulting from that access. Bank of America also said it is “excited about what AI responsibly used can deliver,” according to the email text reported by TechCrunch.
The warning concerns a basic distinction: an agent using a customer’s password can appear to the bank as the customer—or as someone who has stolen the password. That can make it difficult for a bank to identify whether an instruction was authorized, and may leave customers dealing with an unwanted transaction or an account lock. Axios reported on October 22, citing unnamed sources, that major U.S. banks were finding third-party AI agents in customer accounts, often using the customer’s own login credentials. The report said banks had difficulty determining whether customers intended the activity.
There are also financial risks if an agent takes an action the customer did not expect. U.S. Regulation E excludes certain transfers initiated by a person given access to an account by the consumer from its definition of an unauthorized electronic fund transfer. Whether that provision applies to an AI agent or its operator has not been resolved by a regulator or court, according to the source report. In the European Union, payment rules require customers to take reasonable steps to protect personalized security credentials; how those rules apply to giving credentials to an AI agent has not been tested in court, the source says.
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 Bank Logins Give Agents Too Much Access
A banking password is typically a broad access key, not permission for one narrowly defined task. If an agent logs in with the same credentials a customer uses, it may be able to view information or initiate actions beyond the specific job the customer intended. The bank may also have fewer ways to distinguish the agent’s activity from the customer’s own actions.
The issue matters as AI agents move from answering questions to carrying out tasks involving money. The source reports that Visa told investors trust is a major barrier to agents making payments. It also reports that Robinhood said customers had placed more than $250 million into accounts used by AI agents to trade. Those figures and statements illustrate activity and concerns cited by the source; they do not establish that all such tools are unsafe or that every agent has equivalent access.
Security controls do not automatically determine who bears a loss. The source says Robinhood uses a separate account and spending limits but tells customers they assume all risk for trades made by AI agents. Customers therefore need to check both what an agent can do and what the service’s terms say about responsibility before enabling it.
Products Worth Considering
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
Jackery HomePower 3000 Portable Power Station with Two 200W Solar Panels
Apple AirPods Pro 3 Wireless Earbuds
As an affiliate, we earn on qualifying purchases.
Banks Respond to Agent Account Access
The Bank of America email followed reports that financial institutions were encountering AI agents logging into customer accounts with credentials supplied by users. TechCrunch reported the email on October 27 after a customer shared it on Bluesky, and a Bank of America spokesperson confirmed its authenticity to the publication.
AI agents can read material such as web pages, email, invoices and documents while carrying out tasks. The source identifies prompt injection—malicious or misleading instructions embedded in content an agent reads—as a risk highlighted by OWASP’s list of risks for AI applications. That risk is relevant when an agent has access to financial accounts, since outside content could attempt to influence its actions. The source does not establish that such an attack caused the reported bank-account activity.
One alternative to sharing a full login is to grant limited permissions: read-only access, an approval step for each payment, a virtual card with a spending cap, or a token restricted to a particular merchant, amount and time period. A separate account with a limited balance can also restrict potential losses. These are suggested safeguards in the source, not proof that every provider offers them or that they remove all risk.
“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 to TechCrunch
Legal Responsibility Remains Unsettled
It is not yet clear how existing consumer-protection rules will apply when a customer gives banking credentials to an AI agent or to a company operating one. The source notes that no regulator or court has ruled on whether U.S. Regulation E’s provision for access given by a consumer covers an AI agent. It also says courts have not tested whether providing credentials to an agent violates the EU duty to protect security credentials, including the standard concerning gross negligence.
The bank’s email states its position, but it does not by itself determine how a particular dispute or transaction would be handled under the law. The source also does not provide details about the specific agents or account activity reported by banks, the number of affected customers, or whether any reported activity caused losses. Readers should treat the legal discussion as the source author’s interpretation, not legal advice or a court finding.
Check Permissions Before Connecting
Before connecting an agent to a financial account, customers can check the provider’s permissions and terms, especially what actions it can take and who is responsible if it acts unexpectedly. Where available, read-only access, transaction-by-transaction approval, merchant or amount limits, and a separate low-balance account restrict what the agent can do compared with sharing a primary banking password.
Financial institutions and AI providers will need to clarify how agents authenticate, how customers approve transactions, and how disputed activity is handled. The source gives no timeline for new rules or industry standards, and it is not clear whether banks will introduce a common way to grant agents limited access. Until that changes, the central practical question remains whether a tool needs a full login at all.
Key Questions
What did Bank of America tell customers about AI agents?
The bank’s email said customers should not share their usernames or passwords with outside services, including AI agents. It warned that customers who share credentials are responsible for activity resulting from that access.
Does sharing a password mean a customer is always liable for an AI agent’s actions?
That is not settled. The source says the relevant U.S. and EU legal questions have not been resolved by courts or regulators. The bank’s warning is not itself a final legal determination about every transaction.
What is a safer way to let an AI agent handle a financial task?
Where offered, use limited access such as read-only permissions, approval for each payment, a capped virtual card, or an account with a limited balance. These controls reduce the scope of access but do not guarantee that all risks are removed.
What is prompt injection, and why does it matter here?
Prompt injection is an attempt to influence an AI agent through instructions hidden in content it reads, such as a web page or email. If an agent has financial permissions, an outside instruction could try to steer its behavior; the source does not report that this caused the bank activity described.
Source: ThorstenMeyerAI.com
