Bank of America has warned customers not to share their online banking passwords with AI assistants or AI agents, according to reporting by TechCrunch. The bank’s message is simple: if a customer gives an AI system access to their account, the customer may still be responsible for resulting account activity.

While the warning is directed at banking users, it has wider relevance for European businesses. Many companies are now testing AI assistants for admin, finance, customer service and internal workflows. That creates a practical security question: where should automation stop, especially when credentials, payments or sensitive business data are involved?

What happened

Bank of America has publicly reminded customers not to provide banking passwords to AI assistants. The reported warning also states that customers are responsible for activity resulting from their AI agents’ access.

The development does not appear to be a new regulation or a new technical standard. It is a risk warning from a major financial institution as AI assistants become more capable and more widely used for everyday tasks.

The issue is especially relevant because newer AI tools are increasingly positioned as agents that can perform actions on a user’s behalf, not just answer questions. In practice, that can encourage users to hand over credentials or connect tools to business systems without fully understanding the security, liability and compliance implications.

Why it matters for European businesses

For European SMEs and larger organisations alike, the warning underlines a broader governance problem around AI adoption. Businesses may be tempted to use AI tools to log into banking platforms, retrieve account information, process invoices or support payment workflows. But giving an AI assistant direct access to financial systems can create significant risk.

There are several business concerns:

• Credential exposure: entering passwords into third-party AI tools may expose sensitive access details beyond the bank’s intended authentication flow.
• Unclear accountability: if an AI tool initiates an action, businesses may still remain responsible for the outcome under bank terms, internal policy or general security obligations.
• Data protection and confidentiality: finance-related prompts may contain personal data, transaction details or commercially sensitive information.
• Weak internal controls: using a general-purpose AI assistant for banking tasks can bypass established approval processes, segregation of duties and audit trails.
• Third-party risk: many AI tools rely on external providers, plugins or connected services, which can expand the attack surface.

For companies in regulated sectors, or businesses handling payroll, supplier payments or customer refunds, these risks are not theoretical. They affect financial control, compliance, incident response and cyber insurance posture.

Who may be affected

The issue is relevant to a wide range of organisations, but especially:

• SMEs and founders using AI assistants informally for admin or finance tasks without dedicated IT or security review.
• Finance and operations teams exploring automation for payment processing, reconciliation or account monitoring.
• E-commerce businesses managing refunds, payment platforms and high volumes of financial transactions.
• IT and security teams responsible for identity, access control and acceptable-use policies for AI tools.
• Digital agencies and service providers that may access client financial or platform accounts as part of campaign or web operations.

Even when a business does not share a literal password with an AI assistant, similar concerns can arise with browser-based agents, saved credentials, screen access, session tokens or connected automation tools.

What companies should consider

European businesses adopting AI and automation should review whether staff are using AI assistants in ways that create uncontrolled access to financial systems.

• Do not allow direct sharing of banking passwords with AI assistants, chatbots or agent tools.
• Review AI usage policies to cover financial systems, payment tools, accounting platforms and other high-risk environments.
• Use approved integrations carefully rather than informal credential-sharing, and verify what permissions are being granted.
• Apply least-privilege access so automation tools only receive the minimum access necessary for a specific task.
• Strengthen approval controls for payments, refunds and banking changes, including human review for sensitive actions.
• Train employees on the difference between using AI for drafting or analysis and giving AI systems the ability to act inside critical accounts.
• Assess vendor and data risks before connecting AI tools to finance, ERP or banking-related workflows.

More broadly, this warning is a reminder that AI adoption should be treated as an access-management and governance issue, not only as a productivity opportunity. For European businesses, especially SMEs moving quickly with automation, the safest approach is to keep banking credentials and high-risk financial actions outside general-purpose AI tools unless there is a clearly approved, secure and auditable setup.