Mastercard to strengthen trust in agentic commerce

Mastercard has unveiled a new open standard designed to address one of the most pressing challenges in the rise of AI-driven commerce: how to verify that a transaction initiated by an artificial intelligence agent genuinely reflects the consumer’s intent.

Mastercard to strengthen trust in agentic commerce

The initiative is designed to create a shared and auditable source of truth across the payments ecosystem, allowing consumers, merchants and issuers to confirm exactly what was authorised and executed during a transaction.

As AI-powered agents increasingly search for products, negotiate prices and complete purchases autonomously, the question of trust has moved to the centre of payments innovation.

Creating a Cryptographic Audit Trail for AI Transactions

Verifiable Intent links together three elements of an agentic transaction: the consumer’s identity, the instructions provided to the AI agent, and the final outcome of the purchase.

The resulting record is cryptographically secured, ensuring that it cannot be altered once created.

This structure effectively creates an audit trail that can be referenced in the event of disputes, errors or fraudulent activity.

If a transaction is questioned, all parties involved — including the consumer, merchant, issuing bank and payment network — can review the same immutable record to determine whether the AI agent acted within its authorised parameters.

The new framework will be integrated into Mastercard’s Agent Pay Infrastructure, an initiative launched to support transactions initiated by autonomous software agents.

Over the coming months, Verifiable Intent will be incorporated into Agent Pay’s application programming interfaces (APIs) to facilitate adoption among developers and payment partners.

Designed to Work Across Emerging Agentic Protocols

One of the key design features of Verifiable Intent is interoperability. The standard has been built to operate alongside emerging agentic commerce protocols rather than compete with them.

It is intended to work with systems such as the Agent Payments Protocol and the Universal Commerce Protocol, both of which aim to standardise how AI agents interact with merchants and payment services.

Mastercard has based the framework on widely recognised technical specifications developed by organisations including the FIDO Alliance, EMVCo, the Internet Engineering Task Force and the World Wide Web Consortium.

To encourage broad participation, Mastercard has also open-sourced the Verifiable Intent specification and an initial reference implementation on GitHub. Developers, payment platforms and merchants are being invited to review and contribute to the framework.

Industry Support Signals Growing Focus on AI Trust

The initiative has already attracted backing from several major technology companies, including Google, IBM and Checkout.com.

Their involvement highlights how rapidly agentic commerce is moving from conceptual experimentation to real-world deployment.

As AI agents begin to handle purchasing decisions and financial transactions, establishing verifiable trust infrastructure may become as important as speed or convenience.

Mastercard’s proposal reflects a broader industry recognition that automation alone is not sufficient.

In a world where machines increasingly transact on behalf of humans, proving intent — not merely assuming it — could become the cornerstone of the next generation of digital payments.