Mastercard yesterday introduced Verifiable Intent, a new open, standards-based trust layer designed to address one of the most pressing questions in autonomous commerce: how does anyone know that an AI agent buying on a consumer's behalf is actually doing what it was told? Co-developed with Google and endorsed by eight industry partners - IBM, Worldpay, Fiserv, Getnet, Checkout.com, Basis Theory, and Adyen - the initiative creates cryptographic proof of consumer authorization at the moment an AI agent initiates a transaction.

The announcement, dated March 5, 2026, was authored by Pablo Fourez, Mastercard's chief digital officer. It arrives less than two months after Mastercard launched its Agent Pay infrastructure at the National Retail Federation's annual conference on January 11, 2026. That initial launch established the broad architecture for AI-executed payments; Verifiable Intent now adds the authorization and audit layer on top of it.

What the technology actually does

At its technical core, Verifiable Intent links three elements into a single, privacy-preserving record: the identity of the cardholder authorizing the AI agent, the consumer's specific instructions, and the interaction between the agent and the merchant that results in a purchase. According to Mastercard, if a dispute arises, all parties can rely on that audit trail to resolve it quickly rather than relying on conflicting accounts.

The system uses a technique called Selective Disclosure to manage what data gets shared. Only the minimum necessary transaction information is released across parties, and only at the moment it is needed - for fraud mitigation or dispute resolution. Sensitive details beyond that process remain unexposed. According to Mastercard, this approach is designed so that trust and privacy are not tradeoffs.

Technically, Verifiable Intent is built on widely adopted specifications from four standards bodies: the FIDO Alliance, EMVCo, the Internet Engineering Task Force (IETF), and the World Wide Web Consortium (W3C). The choice of these four organizations is deliberate. Each represents a different layer of the digital authentication and web standards ecosystem, and building on them means the specification does not require proprietary infrastructure. According to Mastercard, the system is designed to be protocol agnostic - capable of working across agentic protocols, devices, wallets, platforms, and other payment networks as the ecosystem continues to develop.

Aligned specifically with Google's Agent Payments Protocol (AP2) and Universal Commerce Protocol (UCP), Verifiable Intent creates what Mastercard describes as a tamper-resistant record of what a user authorized when an AI agent acts on their behalf. The UCP was launched by Google on January 11, 2026, establishing open-source technical standards for AI agents to execute purchases across different retail platforms without requiring custom integrations for each merchant.

Over time, Mastercard says Verifiable Intent will be strengthened through integration with Verifiable Credentials, a W3C standard that would make consumer authorization and delegation more explicit, portable, and cryptographically verifiable across systems.

Open-sourcing and developer access

Mastercard today open-sourced the Verifiable Intent specification and an initial reference implementation. Both are available on GitHub and at verifiableintent.dev. Agent platforms, payment enablers, merchants, and developers are invited to review, contribute, and build on the specification. According to Mastercard, API specifications and developer tools for using Verifiable Intent with Mastercard Agent Pay will follow on Mastercard Developers.

The decision to open-source the specification mirrors the approach taken across the broader ecosystem. Google's Universal Commerce Protocol is similarly available through GitHub, with technical specifications defining REST and JSON-RPC transport layers. PPC Land's analysis of UCP noted in January that open-source commerce standards carry the potential to commoditize the interfaces that e-commerce platforms have previously used as competitive moats - a structural dynamic that Verifiable Intent, as an authentication and authorization layer, sits directly on top of.

The integration roadmap

In the coming months, according to Mastercard, Verifiable Intent will be integrated directly into Mastercard Agent Pay's intent APIs. That integration is intended to drive real-world adoption with partners and support scalable agent deployments across platforms. Mastercard also said it will continue working with industry bodies to define complementary standards for conversational AI in commerce.

The road from announcement to integration is not trivial. Agent Pay itself launched only in January 2026. Target and Walmart brought checkout directly into Google's AI assistant on the same day as Agent Pay's debut, representing the first major retail implementations of UCP. Those implementations use existing authentication flows. Adding Verifiable Intent's cryptographic authorization layer on top of live transactions is the next step.

Two scenarios: human in the loop and fully autonomous

Mastercard distinguishes between two operational modes in its announcement. In agent-assisted, human-in-the-looppurchases, Verifiable Intent confirms the consumer was present, approved the shopping cart, and authorized the transaction. Merchants can fulfill these orders without adding friction, according to Mastercard, because the authorization record is already established.

In more autonomous scenarios - where the agent acts independently within pre-set parameters - Verifiable Intent provides what Mastercard calls transparency and agency that allows merchants to assess whether additional confirmation is needed before completing a transaction. This is where the dispute-resolution value becomes most apparent. If an agent books a flight outside a stated budget, or purchases the wrong product variant, the question of what was actually authorized moves from being a matter of memory to one of cryptographic record.

These challenges extend beyond consumer purchases. In machine-to-machine commerce and programmable money implementations - enterprise procurement, cross-border payments, AI agents managing supplier relationships - a consumer or business is always delegating spending authority to a system. According to Mastercard, those systems must make decisions at speed and scale while remaining faithful to the intent of the user.

What industry partners say

The list of endorsing companies spans payment processors, merchant acquirers, and infrastructure providers. Their statements reflect both the opportunity and the underlying concern.

"As AI agents begin to act more independently, it's essential that user intent remains clear, provable, and protected," said Stavan Parikh, vice president and general manager, Payments, at Google. "Strong, interoperable trust infrastructure like Verifiable Intent that is compatible with Agent Payments Protocol is a natural accelerator for scaling agentic commerce, and we're proud to have collaborated with Mastercard on this initiative."

"Over the last couple of months we've seen adoption of AI agents skyrocket," said James Armstead, chief technology officer at Basis Theory. "As these agents start participating directly in commerce - making purchases and completing transactions on behalf of consumers and businesses - the most important question becomes trust. How do you know the intent behind a transaction is real? Mastercard's Verifiable Intent API is an important step toward solving that. By combining Mastercard's network with Basis Theory's tokenization and secure data infrastructure, we can help enable a new model of commerce where agents can initiate and complete payments safely and transparently."

Fiserv framed the integration in terms of merchant outcomes. "Verifiable Intent marks a meaningful step forward in how intent, authentication, and accountability are established in agentic commerce," said Sanjay Saraf, chief product officer, Merchant Solutions, Fiserv. "By integrating with Mastercard Agent Pay, Fiserv is delivering a new trust layer that helps merchants secure agent-initiated transactions. As agents take on greater responsibility, Verifiable Intent enables merchants to proactively reduce fraud, strengthen dispute outcomes, and maintain customer trust."

Adyen's chief technology officer Tom Adams connected the initiative to the broader merchant relationship with consumers. "Merchants have built their businesses on the trust of their customers," Adams said. "As consumers increasingly transact through AI agents, having a verifiable, privacy-preserving way to confirm customer intent becomes foundational. Approaches like Verifiable Intent can help merchants anchor agent-initiated transactions in explicit authorization, while retaining control over identity, data, and the customer experience."

IBM's Kirstin Kirtley Silva emphasized enterprise deployment at scale. "As agent-driven payments mature, we're focused on making them ready for the real world - with strong orchestration, clear policies, and built-in transparency. Verifiable Intent makes user authorization simple and secure, so agents can act safely across platforms," she said. "For merchants, that means every transaction is traceable, and disputes are easier to resolve. We're thrilled to align this with IBM's orchestration layer to help partners move from early prototypes to large-scale adoption."

Worldpay's Reshmi Suresh, head of agentic commerce, noted the open-standard dimension: "Having an open standard that allows them to verify intent, while also giving them the choice to preserve their privacy, is key to increasing adoption. We're excited to partner with Mastercard to evolve this framework of Verifiable Intent, on behalf of our merchants and their consumers."

Why this matters for digital marketing and advertising professionals

The advertising and marketing technology community has been tracking agentic commerce closely since 2025. The shift from human-initiated to agent-initiated transactions carries fundamental implications for how advertising works. If an AI agent completes a purchase based on pre-set parameters, the role of the ad - the impression, the click, the moment of persuasion - may occur at delegation, not at transaction. That changes attribution models, bidding strategies, and the entire concept of the purchase funnel.

Google's introduction of shopping ads in AI Mode on February 11, 2026 represented the first major advertising format designed for agentic discovery. AI Mode at that point had reached over 75 million daily active users, according to Google-commissioned research cited at the time. Those ads appear clearly labeled as sponsored within AI-generated conversations during product discovery moments - a structural acknowledgment that the discovery phase, not the transaction phase, is where advertising now needs to operate.

Verifiable Intent does not directly address advertising. But it matters for marketers for a different reason: it determines whether the underlying agentic commerce infrastructure becomes trusted enough to achieve scale. Without a credible mechanism for proving that AI agents acted within authorized parameters, consumer hesitancy and merchant liability concerns will slow adoption. PPC Land's coverage of AI shopping adoption noted in November 2025 that McKinsey projected the global agentic commerce opportunity could reach $3 trillion to $5 trillion by 2030, with US B2C retail alone accounting for $900 billion to $1 trillion. That scale depends entirely on whether the trust infrastructure keeps pace with the technical capability.

The surveillance pricing controversy that followed Google's UCP launch in January 2026 illustrated precisely the kind of consumer concern that trust infrastructure needs to address. Critics characterized the system as enabling personalized upselling through analysis of chat data. Google denied the claims, citing merchant policies. But the controversy underscored a broader dynamic: as AI systems acquire more purchasing authority, the public scrutiny of that authority intensifies.

Verifiable Intent positions Mastercard - and by extension Google - as infrastructure providers who take that scrutiny seriously. The cryptographic record it creates is not just a fraud-prevention mechanism. It is also a claims mechanism: evidence that the authorization existed, evidence that the instructions were captured, evidence that the transaction matched the mandate. In a world where regulatory bodies are increasingly examining automated decision-making, that kind of audit trail has value beyond the immediate transaction.

The competitive context

Mastercard and Google are not the only parties building agentic commerce infrastructure. OpenAI launched instant checkout capabilities with Stripe on September 29, 2025, introducing the Agentic Commerce Protocol for AI-mediated transactions. Microsoft and OpenAI's positioning for retail AI was documented in April 2025, covering the Copilot Merchant Program and OpenAI's integration of Shopify checkout. Visa has pursued its own agentic payment protocols, and Visa's infrastructure for AI agents was documented in detail in January 2026.

What distinguishes Verifiable Intent is its explicit focus on the authorization record rather than the payment flow itself. AP2, UCP, and the Agentic Commerce Protocol all address how transactions are executed. Verifiable Intent addresses whether those transactions were authorized in the way the consumer intended. That is a narrower but arguably more legally significant question.

Cloudflare, notably, had already begun working with both Visa and Mastercard in October 2025 on security protocols for AI agents, suggesting that the authentication and verification layer of agentic commerce has been recognized as a distinct engineering problem for some time.

Google's UCP checkout help page, published on March 2, 2026 - just three days before today's announcement - specified that merchants must use the native_commerce product attribute to display a Buy button on eligible listings, and that payments use full primary account numbers (FPANs) stored in Google Wallet via Google Pay. That merchant-facing documentation provides the implementation context into which Verifiable Intent's authorization layer will eventually slot.

Timeline

Summary

Who: Mastercard, led by chief digital officer Pablo Fourez, in collaboration with Google and eight industry partners - IBM, Worldpay, Fiserv, Getnet, Checkout.com, Basis Theory, and Adyen.

What: The introduction of Verifiable Intent, an open, standards-based cryptographic trust layer for agentic commerce. The specification creates a tamper-resistant record linking consumer identity, specific purchase instructions, and merchant transaction data into a single privacy-preserving authorization record. It is built on specifications from FIDO Alliance, EMVCo, IETF, and W3C, aligned with Google's Agent Payments Protocol and Universal Commerce Protocol, and uses Selective Disclosure to limit data sharing to the minimum necessary. The specification and an initial reference implementation were open-sourced today at verifiableintent.dev and GitHub.

When: Announced on March 5, 2026. Integration into Mastercard Agent Pay's intent APIs is planned for the coming months. The announcement follows Mastercard's Agent Pay infrastructure launch on January 11, 2026, and Google's simultaneous launch of the Universal Commerce Protocol.

Where: The specification is publicly available at verifiableintent.dev and on GitHub. Integration with Mastercard Agent Pay will be documented on Mastercard Developers. Real-world deployment will occur across platforms and partners using Mastercard's payment network.

Why: As AI agents acquire the ability to execute financial transactions autonomously on behalf of consumers, the existing model of payment authorization - a human tapping a card or clicking "buy" - no longer applies. Without a mechanism for proving what a consumer actually authorized, merchants cannot confidently fulfil agent-initiated orders, issuers cannot distinguish legitimate activity from fraud, and consumers have no recourse if an agent acts outside its mandate. Verifiable Intent is designed to fill that gap, providing a cryptographic audit trail that survives dispute resolution while preserving consumer privacy through selective data disclosure.

Share this article
The link has been copied!