Mastercard is expanding its Agent Pay program with new trust and intelligence services designed to help banks and merchants determine when AI agents make purchases and whether those transactions are legitimate.
Announced on September 30, the new services combine identity, intent, behavioral and fraud information to provide more context around AI-initiated transactions. They form the intelligence layer of Mastercard’s Agent Pay Trust Framework, which covers identity, intent, controls, execution and intelligence.
The development addresses a problem that is becoming increasingly relevant as AI moves from recommending products to purchasing them. Mastercard projects that one in ten consumers will routinely use AI agents to make purchases by 2030.
One of Mastercard’s first new services is a probability score that estimates whether an AI agent initiated a transaction. The service is initially being rolled out for testing in the United States.
Over time, Mastercard plans to strengthen the score with additional information about behavioral patterns, merchant risk, transaction activity, credential risk and consumer behavior.
This additional context could help solve an important payments problem. An AI agent acting on a consumer’s instructions may behave differently from the consumer themselves, potentially making legitimate transactions appear suspicious.
Mastercard gives the example of a consumer asking an AI assistant to plan and book a trip. The agent could purchase flights, accommodation and transportation from several merchants within a short period. Such activity might normally trigger fraud controls, even though the customer authorized every transaction.
By providing more context about the agent and the transaction, Mastercard wants to help banks and merchants distinguish legitimate agent activity from purchases that require further verification.
Mastercard is also working with Cloudflare to explore how web and payment-network signals can provide better visibility into AI-driven transactions while preserving privacy. Another collaboration with Skyfire focuses on “Know Your Agent” technology, which helps businesses identify and verify trusted AI agents.
These developments complement Mastercard’s wider agentic commerce infrastructure. Its Agent Pay program already includes Verifiable Intent, developed with Google, which creates a tamper-resistant record of what a user authorized an AI agent to do.
The company has also introduced Agent Connect and expanded its Agent Suite for Merchants. These services are intended to help merchants connect with external AI agents and make information such as product details, pricing, availability and fulfillment options accessible across AI-powered shopping experiences.
The Mastercard announcement focuses primarily on trust at the transaction stage, but an AI purchase begins much earlier.
Before an agent can make a secure payment, it needs to understand what the consumer wants, discover suitable products and compare their characteristics. Mastercard itself notes that merchants need product information that AI agents can find, understand and act on.
This makes structured product data particularly relevant for platforms such as Icecat. Accurate specifications, identifiers, pricing-related information, product relationships and availability data can help AI systems identify and compare products before a transaction reaches Mastercard’s payment infrastructure.
For example, an agent asked to find a compatible laptop charger within a specific budget needs reliable information about connector type, wattage and compatibility before it can confidently make a selection. Trust at checkout matters, but so does the accuracy of the decision that led to checkout.
Agentic commerce therefore requires several layers to work together. AI agents need reliable product information to make informed choices, clear authorization to act for consumers and payment infrastructure capable of recognizing legitimate automated transactions.
Mastercard’s latest services address one of those crucial layers: making sure that when an AI agent reaches the point of purchase, businesses can understand who is acting, what was authorized and whether the transaction can be trusted.
Read further: News, agentic commerce, AI, e-commerce, ecommerce, Icecat, Mastercard
