🔍 Read the full analysis: Want AI Help With Money? Start With Safer Account Access on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Bank of America warned customers not to share banking credentials with AI tools, saying customers are responsible for activity resulting from access they provide. The report argues that limited permissions, payment approvals and spending caps are safer than giving an agent a full account login, while the legal and financial consequences remain unsettled.
Bank of America warned customers not to give their banking usernames or passwords to AI tools, saying customers who share credentials are responsible for activity resulting from that access. The warning comes as banks confront third-party agents logging in with customer passwords and financial companies explore ways for AI agents to make payments and trades.
The email, titled “Protecting your finances in the AI World,” says banking 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. A customer posted a screenshot on Bluesky, TechCrunch reported on October 27, and a Bank of America spokesperson confirmed the email was authentic to the publication. The bank also said it is “excited about what AI used responsibly can deliver.”
The message puts responsibility for resulting activity on the customer, but it does not set out a new technical access system for AI agents. The source report says that on October 22 Axios reported, citing unnamed sources, that major U.S. banks were finding unauthorized third-party AI agents in customer accounts, often logged in with the customer’s password. Banks reportedly had difficulty distinguishing customer-directed activity from activity the customer did not want.
Meanwhile, Visa told investors this week that trust is a major barrier to AI agents making purchases, according to the source report. Robinhood said customers had put more than $250 million into accounts that AI agents trade. Robinhood also reportedly uses a separate account and spending limits, while warning customers that they “assume all risk” for AI-agent trades. These developments show financial firms and customers are testing agent access even as responsibility and safeguards remain contested.
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 Full Account Access Raises the Stakes
A bank password functions as a broad credential: an agent logging in with it may be able to take actions the account holder can take, rather than perform only the task the customer intended. The bank may see that login as the customer’s activity—or as a potential intrusion—making it harder to tell whether a transfer or other action was authorized. That can expose users to unexpected transactions or account restrictions.
Responsibility for losses is also uncertain. The source article reads U.S. Regulation E as potentially excluding some transfers made by someone given access by the consumer, and points to European rules requiring customers to protect their security credentials. But it explicitly says that applying those provisions to AI agents has not been settled by a regulator or court. The practical warning is clear, while the precise legal outcome in a specific case is not.
The report recommends giving agents narrower permissions instead of a master credential: read-only access where possible, approval before each payment, spending limits, a virtual card, a token scoped to a particular merchant and time period, or a separate account containing only funds set aside for the task. Such controls can limit what an agent can do; they do not, by themselves, decide who bears the loss if something goes wrong.
Products Worth Considering
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
Dell UltraSharp 27 Thunderbolt Hub Monitor (U2726DER)
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Banks Respond to Agent Logins
The developments come as AI systems move from generating recommendations to taking actions across websites and services. For a financial account, that difference matters: an assistant that only summarizes a balance has different access needs from one authorized to initiate purchases or trade assets.
The source report describes prompt injection as a further risk. An agent may read instructions embedded in web pages, email or files and be steered away from the user’s intent. It cites OWASP’s AI application risk list, where prompt injection is identified as a leading risk. That concern reinforces the case for limiting permissions and requiring confirmation for consequential actions, rather than relying on an agent to interpret every instruction correctly.
The report also distinguishes platform safeguards from liability. A separate account or spending cap can reduce potential exposure, but Robinhood’s reported warning that customers assume all risk for agent trades shows that a company’s safety measures do not necessarily mean it accepts responsibility for losses.
““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 email
Who Pays for an Agent’s Mistake?
The legal treatment of AI agents using customer credentials is unsettled. The source report offers its author’s reading of U.S. and EU payment rules, but says no regulator or court has ruled on whether giving credentials to an AI agent falls within the cited provisions. In the EU, it is also unclear whether doing so would meet the legal threshold of gross negligence in a particular case.
The report does not identify the banks involved in the alleged unauthorized agent logins, provide figures for how many accounts were affected, or describe specific resulting losses. It also does not establish which permissions or liability terms will become standard across banks, payment networks and AI services. The Robinhood figure is a reported amount customers placed in accounts used by agents; it is not a measure of trading performance or losses.
Safer Permissions Before Wider Agent Use
Customers considering financial AI tools can first check what access an agent requests and whether it can operate without a banking password. The source report’s practical alternatives include read-only access, per-payment approval, transaction limits and a separate account with a capped balance. Users should also review the service’s terms to understand who is responsible for transactions made by an agent.
For financial firms and agent developers, the next test is whether access can be limited by task, amount, merchant and duration, and whether customers can revoke it. The source material does not give dates for a new industry standard or regulatory decision. Until those questions are answered, the distinction between an agent’s technical safeguards and the customer’s legal responsibility remains central.
Key Questions
What did Bank of America tell customers about AI agents?
It warned customers not to share banking usernames or passwords with outside services, including AI-enabled tools. The email said customers are responsible for activity resulting from credentials they share.
Why is giving an AI agent a banking password risky?
A password may give an agent broad access and make its activity look like the customer’s login. That can make it harder for a bank to distinguish an intended action from fraud or an agent’s mistake.
What are safer alternatives to sharing a password?
The source report suggests read-only permissions, approval before each payment, spending caps, virtual cards, task-specific tokens or a separate account with limited funds. Availability depends on the bank and service.
Are customers legally responsible for every AI-agent transaction?
That is not established by the source material. It discusses possible implications of U.S. and EU rules but says the application of those rules to AI agents has not been tested by a regulator or court.
Does a spending limit mean a company accepts liability?
No. A limit can restrict potential exposure, but it does not settle who is responsible for losses. The report says Robinhood’s customer terms warn users that they assume all risk for trades by AI agents.
Source: ThorstenMeyerAI.com
