Bank of America Warns Customers That Sharing Banking Passwords With AI Agents Voids Fraud Liability
What happened
Bank of America issued a customer-facing email, reported by Bank of America kindly asks that you don't share your banking password with your AI assistant, warning users not to provide banking usernames or passwords to AI agents or tools. The bank listed specific risks: unauthorized transactions, account lockouts, and credential exposure through security flaws in the AI tools themselves. Most significantly for compliance teams, the bank stated explicitly that its account agreements place liability for any resulting activity on the customer who shared the credentials. The warning reflects a broader industry reality: no secure, standardized protocol yet exists for AI agents to access banking systems on a user's behalf without raw credential sharing. This places enterprises in an exposed position if employees or automated finance workflows are delegating account access to AI tools.
Why it matters
- ·Organizations whose employees or finance systems use AI agents to access banking accounts may have unknowingly transferred fraud liability to the enterprise itself, not the bank. Any transaction an AI agent completes using shared credentials is now the organization's legal responsibility under standard account terms.
- ·This warning exposes a gap in enterprise acceptable use policies. Most policies address what data employees may share with AI tools. Few explicitly restrict sharing financial credentials or require that AI agents use delegated access methods rather than passwords. Compliance teams need to close that gap before an incident forces the issue.
- ·Financial services firms watching peer institutions' posture should treat this as an early supervisory signal. Regulators examining AI agent deployments, including banking examiners now asking banks to prove AI agents can be audited, will likely scrutinize whether firms have adequate controls over how agents obtain and use account credentials.
Stay ahead of stories like this
Get every US AI governance development like this one, plus the rest of the week's developments. Every Thursday.
Recent issues
- Regulators, litigants, and attackers are all moving faster than most organizations' AI governance frameworks, shifting leverage away from deployers and toward those who can prove control, accountability, and honest disclosure.21 Oct
- Three separate incidents this week reveal that AI vendors cannot be trusted to self-police, and that harm from deployed agents can run unchecked for months before anyone inside the organization notices.15 Oct
Free every Thursday. Unsubscribe anytime.
Governance controls affected
What to do now
- ☐Ask your finance, treasury, and accounts-payable teams whether any AI tools or automated workflows currently log into banking portals using shared usernames and passwords, and document every instance found.
- ☐Review your enterprise AI acceptable use policy and confirm it explicitly prohibits employees and automated systems from providing banking credentials to AI agents; update the policy if it does not.
- ☐Work with your legal and finance teams to assess whether any existing AI agent access to banking accounts has triggered the liability provisions in your account agreements, and report findings to your CFO and general counsel.
- ☐Contact your banking providers to ask whether they offer delegated access or token-based options that allow AI agents to perform limited, audited actions without receiving a full password, and prioritize migration to those methods.
- ☐Add AI agent credential use to the scope of your next third-party AI vendor due diligence review, specifically asking vendors whether their tools request, store, or transmit banking credentials and under what conditions.
What to watch next
The financial industry has not yet settled on a secure standard for how AI agents should access banking systems without raw credential sharing. Compliance teams should watch for guidance from banking regulators, including the Federal Reserve, OCC, and FDIC. That guidance should address acceptable authentication methods for AI agents acting on behalf of customers or enterprises. Developments around agentic payment frameworks, such as those already under examination at American Express and others, will also shape what controls become baseline expectations. The absence of a regulated standard today does not reduce current liability exposure under existing account agreements.