Mastercard Expands Agent Pay, Launches Agent Trust and Intelligence Suite

Recently, Mastercard announced the expansion of its agentic payments program, Agent Pay, launching a new suite of trust and intelligence services to provide richer context for AI transactions. This suite of services integrates identity verification, intent recognition, behavioral analysis and anti-fraud insights, aiming to help financial institutions and merchants form a unified, clear understanding of AI-driven payment activity, thereby enabling more precise authorization decisions, reducing unnecessary payment friction, and delivering a seamless, smooth transaction experience for consumers. Moreover, this new suite of services also forms the intelligence layer of Mastercard's Agent Pay trust framework, dedicated to laying a solid foundation of trust across the entire agentic commerce chain.

A New Era of Commerce: Every Payment Needs Complete Contextual Support

Today, agentic commerce is rapidly moving into the mainstream: it is predicted that by 2030, one in every 10 consumers will regularly use agents to complete shopping. As the scale of agentic commerce expands, banks and merchants increasingly need clarity on: whether a transaction involves an agent, which operations the agent is authorized to perform, and whether the related activity is genuine and trustworthy. At the same time, regardless of which payment method consumers choose or which agent they use, every transaction must meet consumers' urgent expectations for security.

"Agents have the ability to make commerce more intuitive, efficient and personalized, but only if individuals and businesses can fully trust these systems to act on their behalf," said Ann Johnson, Executive Vice President, Security Solutions at Mastercard. "By adding agent-specific intelligence and risk insights to every transaction, we want consumers to be able to pay with confidence and confirm transactions with peace of mind, no matter which payment method they choose."

Currently, the first service in the suite—the AI transaction probability score—has already been launched for testing in the United States. This score is used to determine the likelihood that a transaction was initiated by an agent, helping issuers approve AI transactions with greater confidence. In the future, Mastercard will continue to integrate multi-dimensional data such as behavioral characteristics, merchant risk, transaction patterns, credential risk and consumer behavioral tendencies to further optimize the score and more clearly identify AI-driven payment activity throughout the transaction lifecycle.

In addition to the probability score, the newly launched agent trust and intelligence signal system can also help every part of the industry ecosystem make more comprehensive and consistent judgments about the actual circumstances behind a transaction around three core questions:

1. Is this transaction consistent with expected behavior or patterns?

2. Are there anomalies in the agent, merchant, payment credential or transaction itself that require further review?

3. Can the transaction be approved directly, or does it require additional verification?

For example, suppose a consumer wants to use a new AI assistant to plan and book a trip. The agent will automatically filter the most suitable flight, hotel and transportation options, and can complete all bookings with just one authorization. These purchases across multiple merchants may appear anomalous on the surface, but insights from Mastercard can provide banks and merchants with the context needed to approve the relevant transactions, giving consumers a smooth experience from trip planning through booking and payment; however, if signs of risk are detected, the system will flag the transaction before the consumer completes payment and trigger additional review.

Industry-Wide Collaboration: Bringing Trust to Agentic Commerce

As agentic commerce continues to evolve, building trust has become a shared issue for the entire industry ecosystem. Currently, Mastercard is working with partners such as Cloudflare to build a privacy computing environment that can effectively identify AI-driven payment activity while fully protecting data security. In addition, the two parties are also exploring how to combine internet signals with payment network signals to help financial institutions, merchants and platforms improve the transparency of AI transactions.

At the same time, Mastercard is also working in depth with Skyfire—a "Know Your Agent" (KYA) technology service provider for agent identity, verification and payment scenarios. The two parties will work together to help financial institutions and merchants identify trusted agents in the industry ecosystem, make more evidence-based authorization decisions, and deliver a smoother transaction experience for consumers.

Partner Perspectives

"As agents play an increasingly important role in commerce, having a clearer grasp of the full process of transaction initiation, authorization and execution will help better meet consumers' expectations for experience," said Todd Kennedy, Executive Vice President, Payment Strategy at Capital One. "We look forward to continuing to work with all parties in the industry ecosystem to safeguard the security of consumers and merchants."

"Trust determines whether agentic commerce can go the distance," said Stephanie Cohen, Chief Strategy Officer at Cloudflare. "With Cloudflare's network, we can gain deep insight into how agents interact with the internet; by combining that with Mastercard's payment intelligence in a privacy-preserving environment, we can provide financial institutions and merchants with the transaction context they need to efficiently approve AI transactions."

"Every agent that completes a transaction on behalf of a consumer should be identifiable, accountable and auditable," said Amir Sarhangi, Chief Executive Officer and Co-Founder of Skyfire. "And that is precisely the cornerstone of KYA. By extending KYA capabilities to Mastercard Agent Pay, we are committed to providing financial institutions and merchants with a trusted way to identify and verify agents in the industry ecosystem, helping agentic commerce scale safely and confidently."

"As agents increasingly act on behalf of consumers, 'agent identification' has become an essential capability for anti-fraud and identity verification systems," said Suzanne Sando, Principal Analyst, Fraud Management at Javelin Strategy & Research. "Our systems must evolve—not only to identify 'who is transacting,' but also to confirm whether the agent is trustworthy, whether it has been properly authorized, and whether it is acting within the boundaries of consumer intent. Our ultimate goal is to clearly distinguish genuine, authorized activity from automated behavior designed for fraud and theft."