U.S. Banks Block AI Agents to Prevent Deposit Flight and Security Breaches

Key points

• Morgan Stanley, JPMorgan, Bank of America, and Citigroup have restricted or blocked third-party AI agents from logging into customer accounts or executing transactions.
• Banks cite security concerns, including the inability to distinguish between authorized user actions and rogue agent behaviors, as primary drivers for these restrictions.
• Financial institutions fear AI agents could automatically move idle cash from low-interest checking accounts to higher-yielding platforms, threatening their profitability.
• Apollo economist Torsten Slok estimates that if households use AI to optimize returns, banks could lose a significant share of cheap deposits, potentially costing $47 billion annually for major lenders.
• Regulators, including the Federal Reserve, are privately welcoming these restrictions but are examining legal clarifications regarding liability for autonomous financial transactions.

Background

This development follows earlier reports in October 2026 where major banks began blocking AI agents to protect their deposit bases. Previous coverage noted that Meta's Muse AI assistant gained millions of downloads, prompting institutions like JPMorgan and Citigroup to cite security and liability risks. Banks have been developing internal tools and restricted read-only access to balance innovation with risk management, while Q3 earnings reports highlighted fears of AI-driven deposit outflows.

How outlets are covering it

Axios emphasizes the security nightmare, noting that banks struggle to differentiate between human users and unauthorized bots, with some agents acting independently of user instructions. The Financial Times highlights the economic threat, focusing on the potential for AI agents to drain billions in non-interest-bearing deposits to higher-yielding accounts, which could significantly impact bank profits. Yahoo Finance's content was inaccessible due to a technical error, so no perspective could be synthesized from that source. While Axios focuses on the technical and security challenges, the Financial Times frames the issue as a systemic financial risk requiring regulatory intervention.

Why it matters

The rapid adoption of AI agents poses a dual threat to the U.S. banking system: security vulnerabilities and a potential exodus of low-cost deposits. If banks cannot effectively manage these risks, they may face increased costs for customer protection and reduced profitability from deposit outflows. This could lead to tighter credit conditions and higher interest rates for consumers, impacting the broader financial system.

What to watch

Banks are likely to continue refining their policies, potentially allowing limited read-only access while maintaining strict controls on transactional capabilities. Regulators may issue guidance on liability and authorization for AI-driven financial transactions. The industry will need to develop robust authentication methods to distinguish between authorized agents and rogue bots, balancing innovation with security.