The news: Banco Santander and Mastercard have successfully executed Europe's first end-to-end payment initiated and completed by an artificial intelligence system without human intervention, per AI News.

The transaction—which was processed on Santander's live banking infrastructure using Mastercard's Agent Pay framework—marks a significant step toward wider AI-enabled financial operations.

• The pilot was carried out under strict security, governance, and compliance rules within Santander's live infrastructure, though it is not yet available for public use.
• The AI agent worked within predefined customer and bank limits to confirm autonomous systems can initiate, authorize, and complete transactions while meeting standard legal and operational banking requirements.

Zooming in: The Santander-Mastercard test demonstrates that agentic systems can operate within existing regulatory frameworks rather than requiring new ones. 

Mastercard's fraud-scoring systems already analyze nearly 160 billion transactions annually—the scale where agentic systems could eventually operate.

By registering the AI agent with identity credentials, spending limits, and audit trails, the pilot proved autonomous systems can meet the same compliance checks as human transactions.

However, moving from a controlled pilot to millions of daily autonomous transactions introduces exponential complexity and risk. Each AI agent would need unique identity verification, real-time fraud scoring, and clear liability assignment when errors or exploits occur. 

Mastercard recently launched Verifiable Intent, an open standards-based trust layer to help close this gap. This layer creates a record of users’ authorized actions to their AI agents for every transaction, giving cryptographic proof of consumers’ intent to merchants, consumers, and issuers. Verifiable Intent can help prevent agentic fraud by having a record of true permissioned agentic behavior.

Zooming out: Mastercard’s successful pilot of an agentic payment on its Agent Pay signals a new burgeoning power dynamic in payments: protocol dominance. 

As Visa, Mastercard, Google, and others develop the new toll roads of agentic payments, these major players now have to balance the need for interoperability—the ability to work with other competitors’ protocols within a network to facilitate a payment—with total ecosystem control, where one payment provider has the ability to claim maximum share of agentic payments running through its framework: 

We forecast that by 2029, US AI platform-driven ecommerce will drive $144.25 billion in sales.

For Mastercard specifically, its pilot’s success also suggests that the payment network is closer to preserving its duopoly it holds with traditional rails for the next era of agentic commerce. 

What this means for marketers: Agentic payments will shift how consumers interact with brands. When AI agents can autonomously initiate transactions, the traditional purchase funnel—awareness, consideration, decision—shifts into machine-to-machine commerce.

• Brands should develop clear interfaces where customers set spending limits, merchant preferences, and approval rules for agent-initiated purchases. 
• Trust will depend on transparency about what AI agents do and how customers maintain control.

Editor’s note: This story was updated to include information about Mastercard’s Verifiable Intent.