Several prominent US banks and brokerages — among them JPMorgan Chase and Charles Schwab — have put controls in place to block AI-driven automated tools that attempt to log into customer accounts without authorisation, pointing to concerns over both security vulnerabilities and potential financial exposure.

Automation within the tech sector is pushing forward tools that act on a user's behalf — fielding financial decisions and identifying higher-yielding places to park savings. Financial institutions have raised objections to third-party software being granted access to account data and the authority to execute transactions on their behalf.

JPMorgan Chase's position, set out by chief executive Jamie Dimon, draws a line at allowing AI agents to reach into customer accounts without prior authorisation — a stance tied to the liability that arrangement carries for the bank and the people whose funds it holds. One area drawing the bank's attention involves outside software that gains entry to accounts by relying on customers' own login credentials. In a statement addressing data-access methods, JPMorgan's chief financial officer Jeremy Barnum raised concerns that credential sharing and screen scraping expose customers' financial information to unacceptable security risks.

Schwab's stance on the matter mirrors that approach. According to Schwab president Rick Wurster, the brokerage's systems are set up to prevent third-party agents from accessing client accounts without authorisation. According to the firm's position, those limitations exist to safeguard both its own infrastructure and the people who use its services.

Financial institutions and investment firms are tracking a scenario in which AI-driven tools autonomously shift funds out of low-yield deposit accounts and into products carrying higher returns. Dormant balances across deposit accounts lend themselves to automated review, where software-driven tools scan available rates and shift funds toward whichever institution offers a more favourable return. Widespread uptake of these tools by account holders would put pressure on the deposit base that banks draw on to extend credit.

At this point, the question is whether the technology will gain broad adoption — not whether it already has. At a financial industry gathering, Wells Fargo's chief executive Charlie Scharf raised doubts about how readily consumers would hand over broad financial control to autonomous agents.

Banks and other lending institutions are testing narrowly defined applications of artificial intelligence even as they place limits on external providers. A tool called Charley — built to field customer queries and handle a narrow set of account-management tasks such as updating personal details — has been rolled out by Schwab. Pilot programmes are under way at the brokerage, putting AI-driven tools into the hands of financial advisers and a subset of traders.

Efforts at JPMorgan centre on delivering artificial-intelligence-driven offerings without surrendering oversight of account access or the personal data held on customers. A tension has taken shape between the ease that artificial intelligence brings to everyday tasks and the need to keep financial data protected.

Autonomous agents designed to handle personal finances on a user's behalf show promise for streamlining money management, though open questions around mistaken transactions, actions taken without explicit approval, and who bears responsibility when things go wrong remain unresolved. Proposals under discussion would give banks their own governance structures for AI agents — formal systems specifying what those tools are authorised to do and logging whether those boundaries are respected.