Agentic Commerce
Agentic commerce is commerce in which AI agents act on behalf of consumers to initiate or assist with purchases. Mastercard describes its Agent Pay program as supporting payments initiated by AI agents on consumers’ behalf.1 As these transactions become more common, payment providers must help merchants and banks distinguish legitimate agent-led purchases from unexpected or fraudulent activity.1
Trust and transaction assessment
Mastercard has expanded Agent Pay with trust and intelligence services that combine identity, intent, behavioral, and fraud signals to provide context for AI-driven transactions.1 The goal is to help banks and merchants make authorization decisions while reducing unnecessary friction for legitimate purchases.1 A shared intelligence layer is intended to help participants assess whether an activity matches expected behavior, identify anomalies, and decide whether to approve or review a transaction.1
One component is a probability score estimating whether a transaction was initiated by an AI agent.1 Mastercard says the score is initially being tested in the United States, and that it may be enriched with information about behavior, merchant risk, transaction patterns, credential risk, and consumer propensity.1 This type of signal can inform an authorization decision, but it does not itself establish that a transaction is safe or that an agent has the consumer’s authorization.1
Ecosystem and adoption
Mastercard is collaborating with Cloudflare and Skyfire on visibility, verification, and accountability for AI agents.1 The partnerships reflect the need for multiple ecosystem participants to establish trust around agents and their payment activity.1 Mastercard also cites a projection that one in ten consumers will use agents for purchases by 2030, underscoring the anticipated importance of agent trust in commerce.1
Practical considerations
Agentic commerce connects payment authorization with questions of identity, consumer intent, behavioral consistency, and fraud risk.1 Shared signals may improve the context available to issuers and merchants, while the quality of decisions will depend on how those signals are produced and interpreted.1 The source describes Mastercard’s announced approach and intended benefits; it does not provide measured outcomes from deployment or a general standard for verifying agent authority.1
Related Pages
- Financial Market Infrastructures — Agentic commerce introduces AI-related risks to payment trust, complementing this page’s focus on AI, cyber threats, and operational resilience in financial infrastructure.