In September 2026, three of the world's largest payment networks—Visa, Mastercard, and Ant International—announced something that didn't make headlines in the marketing press, but quietly reshaped the roadmap for AI-driven commerce. They launched a collaborative effort to develop a "Know Your Agent" (KYA) interoperability framework, designed to identify and verify AI agents acting on behalf of consumers.
This is not a product launch. It is infrastructure—the same kind of trust layer that made e-commerce possible in the 1990s, when Visa and Mastercard jointly created the SET (Secure Electronic Transaction) protocol. The parallel is not accidental.
At Vivid Visuals, we have been tracking the agentic commerce space closely since our analysis of Anthropic's shopping agent blueprint, where we identified trust as the real bottleneck—not technical capability. The KYA framework is the first concrete, industry-wide response to that bottleneck.
What Happened: A Cross-Network Trust Standard
On September 10, 2026, Ant International, Mastercard, and Visa announced a joint initiative to develop common standards for identifying and verifying AI agents in payment ecosystems. As Reuters reported, the three companies have begun collaboration on a Know-Your-Agent interoperability framework.
The timing is not coincidence. It comes with fresh data on consumer readiness:
- Visa's Trust Index research, released in September 2026, found that 72% of US consumers have used an AI assistant in their payment journey—a dramatic figure that reframes "agentic commerce" from a future concept to a present behavior.
- The same research indicated consumer confidence, not technical capability, will decide how quickly shoppers let AI agents make purchases on their behalf.
- Mastercard's Signals report on agentic commerce had already identified intent, consent, and control as the three pillars required for trust in agent-initiated payments.
The KYA framework is the infrastructure attempt to build those pillars into payment networks themselves—for every transaction, not just select pilots.
Why It Matters: The Missing Layer for Agent-Initiated Payments
To understand why the KYA framework matters, consider what happens today when an AI agent attempts to execute a purchase. The payment network sees a request, but has no standard way to answer three critical questions:
-
Is this request actually initiated by an AI agent, or is it a spoofed attempt? Payment fraud prevention systems were designed for human-initiated transactions, with behavioral signals like card-not-present indicators, device fingerprints, and velocity checks. An autonomous agent produces none of these. It sends a clean API request with no device, no geolocation, no human hesitation pattern.
-
Who authorized this agent to act, and for what scope? A consumer may authorize an AI assistant to search for flights within a budget, but not to purchase a $3,000 airline ticket. Without a standardized convention layer, merchants and payment processors cannot distinguish an agent with narrow purchase authority from one with broad cart-blanché.
-
Who bears liability when the agent makes an error? If an AI purchasing agent overpays for inventory due to a stale price feed, operations not designed for AI can leave the merchant, the agent provider, and the network locked in a three-way dispute. The original Ad Age investigation into advertising to AI agents noted legal accountability as one of the most significant unresolved questions in agentic commerce.
The KYA framework addresses these problems directly: it proposes common standards for agent identity (who the agent is), convention and scope (what it is authorized to do), and traceability (when it acts, remembering these three dimensions).
The Parallel to 3-D Secure: When Risk Infrastructure Unlocks Market Growth
There is a historical precedent for this type of cross-network collaboration. In the late 1990s and early 2000s, online payments were slow to scale because card-not-present transactions carried high chargeback risk. The solution was 3-D Secure—a protocol jointly developed by Visa (as "Verified by Visa") and Mastercard (as "SecureCode"). It added an authentication step that reduced chargebacks, changed the liability shift, and made commerce safe enough to scale rapidly.
The KYA framework is the agentic equivalent. By creating a standard for agent identity and authorization scope, it could shift liability in agent-initiated transactions, reduce fraud risk, and give payment networks confidence to route agent transactions through existing rails.
Whether the market adopts K-YA as rapidly as 3-D Secure depends on two factors: technical simplicity (integrating agent identity into existing payment flows must be trivial for merchants) and economic incentives (the framework must offer clear fraud-reduction value, not just compliance overhead).
Who It Affects: Payment Leaders, E-Commerce Brands, and AI Platform Builders
The KYA framework will impact several groups:
- Payment and Issuing Banks: Banks that sit between consumers and merchants must implement the KYA verification steps to stay in flow. Laggards, as The Financial Brand noted, will find themselves "hopelessly behind," especially on the business banking side where automated vendor payments are already common.
- E-Commerce Merchants: Retailers must prepare for agent-initiated transactions on their checkout APIs. This is not theoretical; marketplaces like Amazon and booking platforms like Expedia are already designing agent-accessible endpoints. The cost of preparation is low (mostly API documentation and authentication updates), but the value of early readiness is high: being first in agent-discoverable catalogs is a market position.
- AI Platform Builders and Founders: If you are building AI agents on platforms like Base44, ensuring your agent can present verified KYA credentials when initiating a purchase will become a baseline requirement—not a differentiator. The platforms that make KYA compliance trivial for their developers will have a meaningful adoption advantage.
Practical Implications: 4 Steps Brands Can Take Now
- Map your current checkout/data API stack. Identify which endpoints an AI agent might interact with, and what authentication each one supports. This is the starting point for all KYA preparation.
- Add machine-readable product/pricing schema (Schema.org Offer). This makes your inventory discoverable to both human searchers and AI agents. It's the foundation. Our guide to tracking your AI search visibility covers the measurement side.
- Champion standardized agent identity frameworks internally. Financial services companies, in particular, should set up compliance teams monitor the KYA standards as they evolve—just as EI DSS compliance became standard practice, KYA compliance may be near.
- Evaluate your agent stacks' KYA readiness. If you are building agents that transact, your internal capacity to present agent identity, scope, and authorization becomes a table stakes. Document this now, before external frameworks force it on you.
The Business Case: Trust Is Not Friction—It Is the Enabler
When we discuss agentic commerce with clients at Vivid Visuals, the most common objection is that "adding trust infrastructure is friction—agents are supposed to remove friction." This misreads how trust works in payments. Authentication steps do not add friction; they add the kind of confidence that makes larger transactions, higher volumes, and broader use cases possible.
Without a trust standard, agent-initiated payments stay constrained to small, safe pilots—$20 orders with single-item checkout, no errors, with human fallback. With a trust standard, they can scale to enterprise-class vendor payments, agent-managed inventory, and customized purchasing authority—all without chargeback risk. Trust infrastructure expands what is possible.
Read next from the blog : How to Track Your Brand's Visibility in AI Search: The AEO Measurement Playbook · How AI-Native Companies Turn Workflows Into Operating Capability · AI Agent ROI Tripled: What Salesforce's Agentic Enterprise Index Reveals · Anthropic's Blueprint for AI Shopping Agents: Why Trust Is the Real Bottleneck