Banks Struggle to Verify Autonomous AI Agents in Payments

Financial institutions have spent decades refining their customer verification processes. They are now facing a new challenge: identifying and verifying the artificial intelligence agents that act on behalf of their clients. Zhuoqun Bian, president of Ant Digital Technologies, warned at the Fortune Leaders Forum in Macau that traditional rules are insufficient for this new reality. In the emerging agent economy, banks must determine who owns the agent, who authorized its actions, and what its specific permissions are.

The scale of this shift is significant. A January report from McKinsey estimates that AI agents could orchestrate up to $5 trillion in global consumer spending by 2030. However, the current digital infrastructure was designed for human interaction. Bian noted that existing systems need to be rebuilt or significantly enhanced to accommodate agents that operate without direct human oversight for every single transaction.

New frameworks for agent verification

To address this gap, major industry players are collaborating on new standards. Ant International, the global payments arm of Ant Group, announced in September that it is working with Mastercard and Visa on a "know your agent" interoperability framework. This initiative aims to allow card networks, digital wallets, and marketplaces to recognize trusted AI agents across different ecosystems. The collaboration is taking place through BuildFin.ai, an industry platform convened by the Monetary Authority of Singapore.

Conflicts between adaptive AI and fixed systems

A fundamental technical conflict exists between how AI agents operate and how payment systems function. An April note from the International Monetary Fund highlighted that AI agents are probabilistic and adaptive. The same prompt can yield different results each time. In contrast, payment systems rely on predictable rules and legal certainty. Systems like card networks and real-time gross settlement require consistent, identical responses to ensure financial stability and trust.

Operational risks outweigh technical failures

Benson Wong, managing director and head of digital at JPMorgan Private Bank, suggested that the primary risk is not technical failure but operational misalignment. He has not encountered cases where agent technology itself failed, but rather situations where the operating model and compliance controls were inadequate. Wong emphasized that the cost of errors scales with the level of autonomy granted to the agent. A minor error in an information query is embarrassing, but incorrect agentic workflows can have vastly scaled financial impacts.

The consensus among experts is that banks must understand their agents more deeply than they currently understand their human customers. As reported by GN technics/ai (en-US), the industry is moving toward a model where verifying the agent's authority and intent is as critical as verifying the customer's identity. This shift requires a fundamental rethinking of compliance and control structures to ensure that autonomous spending remains secure and accountable.