🔍 Read the full analysis: AI Money Tasks Made Safer: Keep Your Bank Password Private on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Bank of America has warned customers not to share their banking usernames or passwords with AI tools, saying customers are responsible for activity that follows such access. The warning comes as banks report concerns about third-party agents entering accounts and financial firms explore agent-led payments; the legal treatment of AI-directed transactions remains unsettled.
Bank of America has warned customers not to share their usernames or passwords with AI agents, saying they are responsible for activity resulting from that access. The bank’s email, confirmed as authentic by a spokesperson to TechCrunch, arrives as financial firms grapple with AI tools entering customer accounts and making payments or trades.
The email, titled “Protecting your finances in the AI World,” says banking credentials should not 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 do share credentials are responsible for activity resulting from that access. The bank also says it is “excited about what AI used responsibly can deliver.”
The warning addresses a practical problem: an agent that signs in with a customer’s password may appear to the bank to be the customer, making it difficult to distinguish an authorized instruction from account misuse. The source material says an Axios report on October 22, citing unnamed sources, described banks finding third-party AI agents in customer accounts, commonly using credentials supplied by account holders. Banks were reportedly blocking some activity and revising terms, but the report’s underlying claims are attributed to those unnamed sources.
Financial companies are also moving toward agent-led money activity. The source says Visa told investors that trust is a major barrier to AI agents paying for purchases, while Robinhood said customers had placed more than $250 million into accounts used by AI agents to trade. That figure is a reported total; the source gives no comparison baseline or precise measurement period. Robinhood’s separate accounts and spending limits do not remove the risk: the source quotes the company as saying customers “assume all risk” for AI-agent trades.
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 Password Access Raises the Stakes
A password is not a narrowly limited permission. It can give an agent access to the same account functions available to the customer, while leaving the bank with little information about which actions the customer approved. A mistake, an agent following misleading instructions, or a compromised tool could therefore affect funds beyond the specific task the customer intended.
Responsibility after a disputed transaction is also unsettled. In the United States, Regulation E excludes from its definition of an unauthorized electronic transfer one initiated by a person who was given the account access device by the consumer, subject to provisions including notice that the person is no longer authorized. The source author says applying that rule to an AI agent or its operator is an interpretation, not a determination by a regulator or court.
In the European Union, the source points to PSD2 requirements for customers to take reasonable steps to keep personalised security credentials safe, with potential liability in cases of intent or gross negligence. Whether giving credentials to an AI agent meets that standard has not, according to the source, been tested in court. These differences mean consumers should not assume that ordinary fraud protections will apply in the same way when they have deliberately shared login details.
Products Worth Considering
Bose QuietComfort Wireless Bluetooth Headphones with Active Noise Cancellation, Mic, USB‑C Charging, Deep Bass, 24‑Hour Battery – White Smok
DJI Osmo Pocket 3 Filter Bundle – Vlogging Camera with 1 CMOS Sensor and 4K/120fps
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Banks Face Agent Access Questions
The bank’s email is part of a broader debate over how financial services should handle software that can act on a person’s behalf. An agent may read pages, emails, invoices, or documents while carrying out a task. The source identifies prompt injection—instructions embedded in material an agent reads—as a way an agent could be steered away from the user’s intent. The source describes this as a leading risk in OWASP’s AI application risk list; that risk does not mean every agent will be manipulated, but it complicates the idea that an agent can safely use unrestricted account access.
The article’s proposed alternatives are permissions with tighter limits: read-only account access, requiring approval for each payment, a capped virtual card, or a scoped token limited to a merchant, amount, and time window. Another option is a separate account or pool of funds with a limited balance. These are presented as safer design patterns, not as guarantees that a transaction will be error-free or that a customer will have no liability.
The key distinction is between authentication and permission. A customer’s main password authenticates access to a broad account; a restricted credential can authorize a defined task. Financial firms are still working out how such controls should operate and how responsibility is allocated when an agent acts incorrectly.
““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 is identified in the source material on whether giving a bank password to an AI agent counts as authorizing its transactions under U.S. rules, or as failing to protect credentials under EU rules. The legal discussion in the source is the author’s reading of the cited provisions, not legal advice or a formal agency position.
It is also unclear how widely banks are finding AI agents inside customer accounts. The cited Axios report relied on unnamed sources, and the source does not provide a confirmed count of affected customers, banks, or disputed transactions. Nor does it establish the terms or loss protections attached to every agent-related account or payment product.
Finally, the source does not specify what security checks the reported Robinhood accounts use beyond separate accounts and spending limits, or whether the reported $250 million refers to deposits, assets held, or another measure. Readers should not treat that figure as a measure of profits, losses, or transaction volume.
Use Limited Access Before Connecting
Customers considering an AI service that can interact with money should check what permissions it requests, what transactions it can make, and who is responsible if something goes wrong. Where available, read-only access, per-transaction approval, spending caps, and a separate low-balance account offer narrower access than handing over a primary banking password. Customers should also review the bank’s and service provider’s terms before changing or disabling approval controls.
For financial firms and developers, the immediate issue is whether agent payments can be authorized with credentials limited to a particular task rather than a customer’s full account. Banks may continue updating access controls and account terms as they assess third-party agents. The source provides no announced deadline for new rules or a resolution of the legal questions, so the scope of customer protection remains an open issue.
Key Questions
What did Bank of America tell customers?
The bank said customers should not share their usernames or passwords with outside services, including AI agents, and warned that they are responsible for activity resulting from shared access.
Does sharing a password mean a customer automatically loses fraud protection?
That is not established by the source. It describes relevant U.S. and EU rules but says how they apply to AI agents has not been settled by a court or regulator. The bank’s warning is not itself a judicial ruling.
What is a safer way to let an AI agent handle a payment?
Use limited permissions where available, such as read-only access, approval for each payment, a capped virtual card, or a token restricted by merchant, amount, and time. These controls can limit exposure but cannot guarantee that an agent will behave correctly.
Why can an AI agent’s access look like account fraud?
If an agent logs in with the customer’s credentials, the bank may not be able to distinguish its activity from the customer’s own actions or from someone who stole the password. That can complicate both transaction review and account access.
What is still unknown about AI agents and banking?
The scale of agent access at banks, the legal treatment of agent-directed transactions, and the protections attached to individual products remain unclear in the source material. It also does not specify a timeline for new industry rules.
Source: ThorstenMeyerAI.com
