JPMorgan and Charles Schwab move to curb unauthorized AI agents as banks worry about customer cash

Banks are increasingly worried that artificial intelligence, when plugged into consumer finance workflows, could change where customer money ends up. In a report published on Yahoo Finance, JPMorgan Chase and Charles Schwab were described as taking steps to block unauthorized AI agents, with the concern that these tools could encourage customers to optimize uninvested cash and reduce the bank’s earnings from holding that money.

The core of the anxiety, according to the report’s framing, is tied to net interest income, the interest a bank earns from the spread between what it receives on assets (including loans and certain cash-like investments) and what it pays on deposits. When customers use automated tools to move balances into arrangements that pay less attention to the bank’s deposit and cash channels, banks can face margin pressure, even if customer relationships remain intact.

For customers, “uninvested cash” typically refers to money sitting idle rather than in active investments. AI-powered “agents” are software systems that can take actions on a user’s behalf, such as analyzing choices and executing steps through connected accounts. If an agent can effectively redirect that idle cash flow, banks could see a shift in deposit behavior and a corresponding impact on income streams that depend on balances staying put or moving through bank-controlled products.

JPMorgan and Schwab’s reported moves to block “unauthorized” agents suggest that the firms are drawing lines around which third-party tools can access or influence customer accounts. The report does not describe the specific technologies, the exact categories of blocked agents, or the enforcement mechanisms in detail. It also does not clarify how the firms distinguish between authorized automation that clients request and automation deemed unauthorized.

While the report emphasizes the link to customer cash, it also points to a broader operational challenge for banks: controlling financial data and account actions in an environment where AI tools can proliferate quickly. Even when the intent is consumer convenience, banks must manage security, compliance, and operational risk, particularly when software can potentially execute actions at scale or outside normal support channels.

In sector terms, the development fits a wider pattern in retail banking, where banks have repeatedly faced competition on “where cash goes,” not just on which bank offers the best loan rates. Interest-bearing cash products, sweep features, money market allocations, and brokerage-linked arrangements can all influence how much money remains in deposits and how much moves into other vehicles. If AI accelerates cash optimization, the speed of balance migration can increase, making income forecasting harder.

Still, the practical details of what JPMorgan and Schwab actually blocked, and how customers can or cannot regain access through sanctioned channels, appear unclear from the Yahoo Finance report alone. The reporting also does not provide quantitative estimates of net interest income exposure or the number of affected agents, so it is not possible to judge how large the financial impact could be based solely on the information described.

What to watch next is whether regulators or industry groups provide clearer guidance on AI agents in financial services, and whether the banks publish more specific policies for authorized agent integrations. Investors and customers will likely look for transparency on approved automation pathways, and whether banks respond with new controls, partnerships, or product updates aimed at retaining cash balances without blocking legitimate tools.