AI Agents Create New Security and Authorization Challenges in Payment Systems
Commercial transactions carried out by artificial intelligence agents give rise to new security requirements regarding the verification of machine authority and data sharing within the financial system.
As the agentic commerce model—where AI agents shop and execute transactions on behalf of humans—becomes widespread, existing payment systems struggle to verify machine authority.
Payment Problems Brought by Agentic Commerce
AI agents already possess the capacity to shop, negotiate, and execute financial transactions on behalf of humans.
However, current payment systems lack a universal authentication method to identify the AI agent behind the transaction and its authorities.
Authorization and Capturing Intent
It is not enough for financial institutions to simply know that a transaction comes from an AI; they need to understand which agent made the transaction and what it is authorized to do.
Expressing the consumer's original intent and matching it with the transaction stands out as a critical phase for future secure payment processes.
Data Fragmentation and Audit Trails
Existing fraud systems are not designed to answer the questions posed by agentic transactions, and the necessary data is divided among different parties.
Data sharing among merchants, AI platforms, and banks is of great importance in terms of the audibility of transactions.
Adding Controlled Friction to Payment Processes
The financial industry, which has eliminated friction in payment steps for decades, is considering intentionally adding some friction for agentic commerce.
Virtual cards and dynamic security credentials are among the solutions explored to make machine authority observable.