Unvetted AI Agents Accessing Bank Accounts: A Growing Security Risk for U.S. Financial Institutions

The discussion examines reports of unvetted, unauthorized third-party AI agents reaching consumer banking accounts at major U.S. financial institutions, and what that means for the security of the financial system. It also considers how banks and the Federal Reserve should balance innovation with cybersecurity, drawing on past incidents involving critical payment infrastructure.

■ AI Agents and Banking Security
- Reports of unapproved AI agents accessing accounts at large U.S. banks
- Examples of agents taking actions without customer authorization
- Banks preparing contingency plans for agents going out of control or being compromised
- The role of APIs connecting fintech companies to the financial system's backbone

■ Systemic Infrastructure Risks
- Key-entity and key-institution risks across BNY Mellon, DTCC, and Fedwire
- A past cybersecurity issue at ICBC and BNY Mellon's use of hard drives for Treasury transactions
- Cybersecurity viewed as an existential risk by major banks and their leadership
- The growing role of central banks in protecting payments and user accounts

This is intended for viewers interested in fintech, cybersecurity, banking regulation, and central bank policy. It provides context on where systemic weaknesses in financial infrastructure may lie and how priorities at institutions like the Federal Reserve could shift over time.