Ant International is joining forces with Visa and Mastercard to develop a new framework for payments conducted by artificial intelligence agents, as the global payments industry prepares for a future in which AI systems could increasingly make purchases and complete transactions on behalf of consumers and businesses.
The collaboration is focused on creating common standards for what the companies describe as “agentic commerce” — transactions in which AI agents can perform commercial tasks with varying levels of autonomy.
As artificial intelligence becomes more deeply integrated into shopping, travel, financial services and everyday digital experiences, payment companies face a fundamental challenge: determining whether an AI agent requesting a transaction is legitimate, authorized and behaving according to the intentions of its user.
Ant International, Visa and Mastercard want to establish infrastructure capable of answering those questions across different payment networks.
At the center of the initiative is the concept of “Know Your Agent,” a framework designed to verify AI agents, connect them with trusted entities and allow merchants and payment providers to recognize legitimate agents across multiple platforms.
The partnership could become an important building block for a rapidly emerging AI-powered commerce ecosystem.
AI Agents Could Influence Trillions of Dollars in Commerce
The potential scale of agentic commerce is enormous.
The companies have pointed to projections from McKinsey estimating that AI agents could manage between $3 trillion and $5 trillion in global consumer commerce by 2030.
AI agents are significantly different from traditional digital assistants.
Instead of simply answering a question or providing a recommendation, increasingly sophisticated agents can potentially complete multiple steps of a task independently.
A consumer could eventually instruct an AI assistant to find the cheapest suitable flight, compare hotels, book accommodation, arrange transportation and complete the necessary payments.
Businesses could use similar technology to order supplies, manage subscriptions, purchase software services or conduct other routine transactions.
That level of automation could dramatically reduce friction in commerce.
But allowing artificial intelligence to control purchasing decisions also introduces significant new risks.
Trust Is the Central Challenge
Jiang-Ming Yang, chief innovation officer at Ant International, has identified trust as one of the most important obstacles facing AI-powered payments.
One major concern is the tendency of artificial intelligence systems to produce incorrect information, commonly described as AI hallucinations.
Hallucinations become considerably more serious when an AI system moves from generating information to taking financial actions.
An inaccurate chatbot response may inconvenience a user. An autonomous agent incorrectly purchasing an expensive product, sending money to the wrong merchant or misunderstanding a spending instruction could produce real financial consequences.
Yang has therefore argued that trust must form the foundation of the transition toward AI-driven commerce.
Payment networks need mechanisms that can determine who controls an AI agent, what permissions the agent has been granted and whether its actions remain consistent with those permissions.
This is where the “Know Your Agent” concept becomes important.
What Is “Know Your Agent”?
The framework has similarities to identity and compliance systems already used throughout financial services.
Banks and financial institutions rely on Know Your Customer, or KYC, procedures to establish the identity of customers and reduce fraud and other financial risks.
Agentic commerce introduces another participant into the transaction: the AI agent.
Payment networks may therefore need a reliable way to establish an agent’s identity and authorization.
A Know-Your-Agent framework could connect an AI system with a verified user, company or service provider while defining what that agent is permitted to do.
The system could also monitor agent behavior.
If an AI agent suddenly begins initiating transactions that are inconsistent with its previous activity or outside the permissions granted by its owner, payment systems could potentially identify the unusual behavior before completing the transaction.
That capability could become particularly important as AI agents become more autonomous.
Interoperability Could Determine Whether Agentic Commerce Scales
Ant International, Visa and Mastercard are not starting from zero.
Each company has spent the past year developing its own protocols and technologies designed to secure AI-assisted payments.
The challenge is making those systems work together.
If every payment network, digital wallet, merchant and AI company creates a completely separate verification system, consumers and businesses could face a fragmented environment.
An AI agent authorized for one payment network might need to complete an entirely new registration process before operating on another.
The partnership aims to prevent that problem.
Yang explained that an AI agent registered through Ant International should ideally be recognizable by Visa or Mastercard without forcing the user to repeat the entire verification process.
That interoperability could significantly simplify the adoption of agentic commerce.
The principle is similar to other areas of digital infrastructure where shared technical standards allow different companies and systems to communicate.
For AI payments, common standards could allow trusted agents to move between merchants, digital wallets and card networks while maintaining consistent identity and authorization information.
Mastercard Emphasizes Consistent Recognition
Pablo Fourez, chief digital officer at Mastercard, has also emphasized the importance of interoperability between Know-Your-Agent systems.
From the perspective of merchants and payment processors, the same trusted agent needs to be recognized consistently across different networks.
Without that consistency, businesses may struggle to determine whether a payment request initiated by an AI system should be accepted.
A common framework could provide merchants with clearer signals.
Instead of treating every AI-generated transaction as an unknown risk, payment infrastructure could verify that the agent is associated with a legitimate user or business and has authorization to initiate the requested purchase.
This could become especially important if AI agents eventually process millions or billions of transactions.
Manual verification would be impossible at that scale.
The trust infrastructure itself will therefore need to be automated.
Visa, Mastercard and Digital Wallets Are Converging
The partnership also reflects major changes in how consumers pay.
Visa and Mastercard remain dominant payment networks across many developed economies, but digital wallets have become increasingly important worldwide.
Their adoption has been particularly strong in markets where consumers moved rapidly from cash toward mobile payments without relying as heavily on physical credit cards.
Ant International operates Alipay+, a cross-border digital payment and merchant services platform that connects with more than 50 electronic wallets globally.
This creates a potentially important bridge between traditional card networks and the rapidly expanding digital-wallet ecosystem.
The distinction between cards and wallets is also becoming less clear.
Many digital wallets are connected directly to credit or debit cards, allowing consumers to use card networks without physically presenting a card.
At the same time, wallets can support alternative payment methods and local financial systems.
Agentic commerce could accelerate this convergence.
An AI agent may ultimately care less about whether a transaction is technically completed through a card, bank account or digital wallet. Instead, it could select the appropriate payment method based on the user’s preferences, cost, rewards or merchant acceptance.
Digital Wallets Already Represent Trillions in Spending
The scale of digital-wallet adoption demonstrates why interoperability matters.
According to payment processor Worldpay, digital wallets represented approximately 56% of global e-commerce transaction value in 2025.
They also accounted for about 33% of transaction value at physical points of sale.
Combined, those payments represented more than $13 trillion in spending.
The growth of wallets means the future of AI commerce cannot be built exclusively around traditional card infrastructure.
AI agents will need to interact with a much broader range of payment systems.
That makes cooperation between companies such as Ant International, Visa and Mastercard strategically significant.
Together, their networks touch a substantial portion of global digital commerce.
Alipay Is Already Experimenting With AI-Powered Purchases
Ant’s experience with AI-assisted payments is not purely theoretical.
Ant International was separated from Hangzhou-based Ant Group roughly three years ago as part of a restructuring that created a business focused heavily on international operations.
Ant Group continues to operate Alipay, one of the dominant digital payment applications in mainland China.
Alipay has recently introduced AI-powered capabilities that demonstrate how agentic commerce could work in everyday situations.
Users can employ AI features to automate repetitive purchasing tasks.
For example, someone could configure an agent to arrange a regular Starbucks order or handle routine ride-hailing requests through Didi.
The AI can initiate the process and prompt the user when payment authorization is required.
These relatively simple examples provide a glimpse into a much broader future.
Instead of repeatedly opening applications, searching for products and manually completing checkout forms, consumers could increasingly communicate their intentions to AI systems.
The agent would handle much of the underlying process.
AI Payments Create New Security Questions
The convenience of agentic commerce also creates a new category of cybersecurity and financial risks.
Payment systems will need to distinguish between legitimate agents and malicious software attempting to impersonate them.
They will also need mechanisms for determining exactly what an agent has permission to purchase.
A consumer might authorize an AI assistant to spend up to $100 per week on groceries, for example, but not allow it to purchase electronics or transfer money.
Businesses may require even more sophisticated controls, including spending limits, approved suppliers and multiple authorization levels.
Another challenge involves accountability.
If an AI agent makes an incorrect purchase, payment providers and merchants will need clear rules for determining responsibility.
Was the transaction properly authorized? Did the AI misunderstand the instruction? Did a merchant provide misleading information? Was the agent compromised?
These questions demonstrate why agentic payments require more than simply giving artificial intelligence access to existing checkout systems.
The underlying payment infrastructure may need to evolve alongside the AI.
A New Layer of Global Payment Infrastructure
The partnership between Ant International, Visa and Mastercard suggests that some of the world’s largest payment companies are preparing for AI agents to become active participants in global commerce.
The opportunity is substantial.
If projections of $3 trillion to $5 trillion in AI-managed consumer commerce by 2030 prove accurate, agentic payments could become one of the most important developments in the payments industry over the remainder of the decade.
But reaching that scale will depend heavily on trust.
Consumers need confidence that AI agents will follow their instructions. Merchants need confidence that agents requesting purchases are legitimate. Payment processors need systems capable of detecting fraud and abnormal behavior.
Interoperable Know-Your-Agent standards could provide part of that foundation.
By allowing verified AI agents to operate across card networks, digital wallets and merchant platforms without repeatedly registering with each system, Ant International, Visa and Mastercard are attempting to create a common trust layer for AI-driven commerce.
The ultimate goal is straightforward even if the technology behind it is complex: an AI agent trusted by one part of the payment ecosystem should be recognizable and trusted throughout the rest of it.
If the companies succeed, consumers may eventually interact with payments very differently.
Instead of choosing a card, opening a wallet or manually entering payment details, users could increasingly tell an AI agent what they want — and allow trusted payment infrastructure to handle everything that follows.
