Mastercard has expanded Agent Pay with trust and intelligence services, working with Cloudflare and Skyfire on AI agent visibility.
The services combine identity, intent, behavioural, and fraud insights to support authorisation decisions and reduce unnecessary friction for consumers.
According to the official press release, the new capabilities form the intelligence layer of the Agent Pay Trust Framework. Mastercard structures the framework around five components: identity, intent, controls, execution, and intelligence.
Why agent-initiated payments need more context
Mastercard frames the expansion around the growth of agentic commerce, citing projections that one in ten consumers will routinely use agents to make purchases by 2030. For banks and merchants, this raises operational questions. They need to know whether an agent is involved in a transaction, what it has been instructed to do, and whether the activity is legitimate. Consumers, meanwhile, expect the same level of security regardless of payment method or the agent they use.
It was also said that AI agents will only change commerce if people and businesses can trust the systems acting on their behalf, while the transaction-level insights are meant to give consumers confidence however they choose to pay.
Probability score enters US testing
The first service is now rolling out for testing in the US. It is a probability score indicating how likely it is that a transaction was initiated by an AI agent, and it is designed to give issuers confidence to approve legitimate agent-led purchases. Mastercard plans to add further intelligence over time on behaviour, merchant risk, transaction patterns, credential risk, and consumer propensity. This will provide a broader view of AI-driven activity across the transaction lifecycle.
Alongside the score, the company is introducing trust and intelligence signals that give ecosystem participants a common view of what sits behind a purchase. The signals address three questions:
- Is the activity consistent with expected behavior or patterns?
- Does the agent, merchant, credential, or transaction show anything unusual that requires further review?
- Should the transaction be approved, or are additional checks needed?
Mastercard uses travel booking to illustrate the use case. An AI assistant asked to plan a trip may book flights, accommodation, and transport across several merchants after a single consumer approval. A bank or merchant could see that cluster of purchases as unusual. The added context is meant to let legitimate activity proceed and to flag risky transactions for review before the consumer completes the purchase.
Partnerships with Cloudflare and Skyfire
Mastercard is building the intelligence layer with ecosystem partners. With the collaboration with Cloudflare, the company will focus on creating privacy-preserving environments to better understand AI-driven payment activity. The two companies are exploring how web and payment network signals can give financial institutions, merchants, and platforms greater visibility into AI-powered transactions.
The company is also working with Skyfire, a provider of KYA technology for AI agent identity, verification, and payments. The partnership aims to help financial institutions and merchants recognise trusted agents across the ecosystem and make more informed authorisation decisions.
The announcement treats transaction-level intelligence, distributed across issuers and merchants, as a core element of trust in agentic commerce. Mastercard is positioning these signals as a shared reference point rather than leaving each party to assess agent activity on its own terms. The company has said it will continue developing this infrastructure with partners. Its stated aim is for AI-powered transactions to be as trusted as any other payment.
