In a speech at Sibos 2026, Federal Reserve Board Governor Christopher Waller discussed how artificial intelligence could improve cross-border payments and enable autonomous commerce, while creating new cybersecurity, authentication, liability and fraud risks. He identified trust frameworks and technical standards as the main requirements for scaling agentic commerce, particularly when AI agents are authorized to make payments without direct buyer involvement. For cross-border payments, large language models could improve sanctions screening and anti-money laundering controls by reducing false positives, while AI agents could optimize payment routing, foreign exchange conversion and liquidity management. Waller expects these models to supplement faster traditional anomaly detection tools. AI could also strengthen cyber defense, but payment operators face an asymmetry because attackers need to exploit only one vulnerability while operators must protect highly interconnected infrastructure. Agentic commerce is likely to begin with consumer transactions in which AI assists with product discovery before progressing toward delegated purchases. Business-to-business payments may also be well suited because agents can operate within supplier and budget rules, although higher transaction values increase the consequences of errors or unauthorized actions. Scaling will require methods to verify an agent’s authority, allocate liability, recalibrate fraud controls for machine behavior and support payments across multiple rails. Waller also highlighted unresolved choices between interoperable and platform-specific standards, and between open systems that accept multiple agents and closed systems tied to proprietary agents.

Original source