Visa, Mastercard and Ant International have launched a joint initiative to develop common standards for identifying and verifying artificial intelligence agents that make purchases on behalf of consumers.
The companies said the effort is focused on creating greater interoperability between payment networks, digital wallets, AI-agent platforms and marketplaces as software agents move from helping people find products to carrying out transactions for them.
The initiative centers on a concept known as Know Your Agent, or KYA. It is intended to provide a common way for different parts of the payment ecosystem to establish an agent’s identity and trustworthiness without requiring every participant to build an entirely separate verification process.
Reuters reported that the three companies announced the initiative on September 10, 2026, as AI systems increasingly gain the ability to conduct transactions for users rather than simply provide recommendations.
In a conventional online purchase, a person typically provides a visible indication of authorization by selecting a product, entering payment information or approving the transaction. AI agents introduce a different process because the consumer may give an instruction in advance and allow software to make decisions and complete the purchase.
That creates new questions for merchants and payment providers. They need ways to establish which agent is making a request, who authorized it, what the agent is allowed to do and whether the transaction should be trusted.
The new initiative does not replace the separate agent-payment systems already being developed by Visa, Mastercard and Ant International. Instead, the companies are seeking common principles that can allow those systems to work across different payment environments while each company retains its own verification, authorization and risk controls.
Visa has already developed its Trusted Agent Protocol, which it introduced in October 2025 with Cloudflare. The protocol is designed to help merchants distinguish legitimate AI agents acting for consumers from malicious automated traffic.
Visa’s technical documentation describes a system in which AI agents can use cryptographic signatures to prove their identity and communicate information associated with a transaction. The protocol is designed to operate over existing web infrastructure and uses the HTTP Message Signatures standard defined by the IETF.
Mastercard has pursued a related but distinct approach through Verifiable Intent, which it introduced with Google in March 2026.
Mastercard’s system is designed to connect the identity of the agent with the consumer’s instructions and the resulting action. The company says it can create a tamper-resistant record of what a user authorized, providing cryptographic evidence that can be used by consumers, merchants and issuers.
Mastercard also says Verifiable Intent uses selective disclosure so that only necessary transaction information is shared when verification is required. The company has open-sourced the specification and an initial reference implementation and says the system is designed to work across agent protocols, devices, wallets, platforms and payment networks.
Ant International is bringing another component to the initiative through its Agentic Mobile Protocol.
Ant introduced the protocol in April 2026 and open-sourced it for use across digital wallets, mobile applications and other connected environments. The company has designed it for AI agents operating through devices and services including smartphones, smartwatches, augmented-reality glasses and in-car systems.
Ant says its protocol includes mechanisms for establishing an agent’s digital identity and authorized capabilities. It also includes an Agent Trust Rating intended to assess an agent’s trustworthiness and influence the level of autonomy it can receive.
The company has additionally developed an agent-to-agent settlement mechanism capable of supporting payments as small as $0.000001, aimed at machine-to-machine transactions and other high-frequency payments.
The three companies are now seeking to make these different approaches more interoperable.
A common KYA framework could allow information about an AI agent’s identity, authorization and trust status to be recognized across multiple payment ecosystems. The companies say this could reduce duplicated identity verification and simplify integration for businesses building agent-based commerce services.
The proposed framework is emerging as companies across the technology and payments industries develop different standards for AI-driven transactions.
Google has developed the Universal Commerce Protocol and its Agent Payments Protocol, while Visa and Mastercard have developed their own systems for agent identity and transaction authorization. These efforts address different parts of the process, making interoperability increasingly important as the market develops.
Mastercard’s research indicates that consumer trust remains a significant challenge. Its August 2026 Signals report found that only 10% of surveyed consumers were willing to let an AI agent complete a purchase autonomously, even as consumers increasingly use AI for product discovery and other shopping activities.
The company’s report also cited a forecast that agent-assisted consumer spending could reach between $3 trillion and $5 trillion by 2030. McKinsey has separately projected that AI agents could orchestrate between $3 trillion and $5 trillion in global consumer commerce by that year.
Those projections help explain why payment companies are working on identity and authorization infrastructure before autonomous purchasing becomes more widespread.
Security is another concern. An AI agent with permission to spend money creates risks that are different from those associated with an AI system that only generates information. Payment providers and merchants need to distinguish legitimate agents from automated systems that may be attempting fraud or unauthorized transactions.
The ability to establish what a consumer actually authorized could also become important when transactions are disputed.
For example, a consumer could instruct an agent to purchase an item within a specified budget, while the agent could potentially select a product outside those restrictions. A verifiable record of the user’s authorization and the agent’s subsequent actions could give payment participants evidence to use when investigating such a dispute.
Mastercard has specifically positioned Verifiable Intent around this issue, linking identity, user instructions and the resulting transaction in a privacy-conscious record.
Ant International has also previously worked with both Visa and Mastercard on agentic payment capabilities. Mastercard lists Ant International among companies participating in its broader Agent Pay ecosystem, while Ant has described its work with both card networks on AI-agent transactions.
Discover more from Aree Blog
Subscribe now to keep reading and get access to the full archive.
