Bank of America limits how AI agents can access customer deposit accounts, aiming to keep money moves under human control

Bank of America is tightening the rules that govern how artificial intelligence agents can interact with customer accounts, according to a report published by Yahoo Finance on Oct. 19, 2026. The change is aimed at keeping “money movements” under human control, a key concern as banks experiment with automation and AI tooling for customer service and back-office operations.

The report says the bank is limiting AI agents’ access to customer accounts and is placing additional restrictions on consumer deposit accounts. Deposit accounts include checking and savings accounts where customers typically hold cash and make everyday transfers, bill payments, and other transactions.

While the post does not lay out a detailed technical plan, the thrust of the announcement is clear: the bank wants to narrow what autonomous systems can do directly in customer accounts. In practice, that means AI may be used more for information retrieval, guidance, or drafting, while actual execution of changes to balances, transfers, and similar actions is kept in the hands of bank employees or the customer, rather than left to software to act on its own.

The move arrives as AI use in financial services spreads across customer support, fraud and risk monitoring, and internal workflow automation. For banks, the operational and regulatory challenge is the same across use cases: minimizing the risk that a system interprets requests incorrectly, acts on incomplete context, or triggers unintended transactions.

Bank of America’s focus on consumer deposit accounts indicates where the bank sees the highest stakes. Deposit products are tightly tied to customer funds and transaction behavior, and they often require strong controls and audit trails. If an AI system can initiate actions directly, banks must ensure that such actions align with customer intent and with internal policies, including limits, approvals, and compliance requirements.

The report also indicates that Bank of America is tightening access “as use of autonomous financial” processes grows. Even when AI is not directly moving money, autonomy can creep in through agentic behavior, where models take multi-step actions to accomplish a goal. The bank’s stated direction suggests it wants to prevent that autonomy from extending to direct execution within accounts.

What Bank of America did not disclose in the Yahoo Finance report is as important as what it did. The post, as described in the available material, does not provide specifics on the exact technical rules, which AI systems are affected, what permissions are restricted, or what alternative workflows the bank will use instead. It also does not detail whether the changes apply to all AI providers, internal tools, or third-party agent platforms, nor does it quantify the operational impact such as changes in response times, customer experience, or costs.

Bank of America also did not, in the information available here, specify whether the tightening is a response to a particular incident, regulatory feedback, or a broader internal review. The bank may still be evaluating how to balance automation with risk controls, but until more details are published, it is not possible to determine whether this is primarily a policy shift, a product change, or both.

Going forward, investors and customers are likely to watch for clearer explanations of the bank’s AI governance approach. Key questions include whether the bank will publish updated guidelines for AI-powered account interactions, how it defines “human control” in day-to-day operations, and whether it will expand or narrow the restrictions over time as AI capabilities evolve and as regulators issue new guidance. For now, The announcement is that the bank is drawing a firmer line between AI assistance and AI execution when customer money is involved.