Mastercard has introduced Agent Pay, a purpose-built payment infrastructure designed to support agentic commerce, where autonomous AI assistants perform transactions on behalf of users. This is a strategic step toward enabling secure, seamless payments in AI-driven environments.

Central to the system is the introduction of Mastercard Agentic Tokens, a new form of tokenization. This is a new class of secure, dynamic tokens derived from the same trusted technology that powers contactless payments, digital wallets, and Payment Passkeys. But now, these tokens are reimagined for machine-to-machine commerce.

Agent Pay is engineered to integrate directly with leading AI platforms. The first implementation will be with Microsoft, connecting Mastercard’s payment network to the Azure OpenAI Service and Copilot Studio. This integration will allow AI systems to securely complete purchases within conversational interfaces, without user intervention at the point of sale.

Key Takeaways

• Mastercard’s new system allows AI agents to securely complete payments on behalf of users using verified digital tokens.
• Each AI assistant receives a unique, traceable token tied to your account, separating real card data from agent activity.
• Initial rollout includes Microsoft’s Azure and Copilot Studio, with additional support from IBM, Braintree, and Checkout.com.
• Mastercard uses tokenization, real-time fraud detection, and identity verification to manage risk and build user trust.

Mastercard Launches Agent Pay for AI-Driven Transactions

Mastercard has introduced something new called Agent Pay. It’s not a futuristic idea – it’s already here, and it changes how people and businesses can let AI handle their purchases. Think of it this way: instead of you shopping online, your digital assistant could do it for you. Not just picking items, but paying for them too.

This system works by giving AI agents a kind of digital payment card that’s tied to your account but doesn’t reveal your real card number. These are called agentic tokens. Each AI assistant gets its token, which makes it easier to track and control what they’re doing. So if your assistant orders a pair of shoes or a business supply, it uses this token to pay, not your real card.

There’s a lot of concern when it comes to letting AI make payments. Mastercard seems to know that and is building checks into the system. Before any payment happens, the AI has to be registered and verified to prove its trustworthiness. This trusted identity is recognized by merchants to differentiate real, authorized agent-initiated transactions from risks.

Let’s say you’re chatting with a shopping assistant online about what to wear to a birthday party. The AI might show you a few dresses and then ask, “Should I go ahead and buy this one for you?” If you say yes, it can finish the purchase using that secure token. You don’t have to open a new tab or manually enter payment info. It’s all done through that conversation.

For businesses, this could be even more useful. Imagine a textile company that needs to order fabric from different suppliers. An AI agent could find the best deals, handle back-and-forth questions, make sure delivery timelines are met, and then pay for it, all through a secure setup without needing someone to check every step.

Mastercard is working with big tech partners to roll this out. Microsoft is one of the main ones. Their tools, like Azure and Copilot Studio, are being built to support these types of AI agents, especially for companies. IBM is also involved, helping larger businesses use AI to automate their internal tasks, including payments.

Payment platforms like Braintree and Checkout.com are joining too. They’ll help merchants accept these agent-driven transactions just like regular payments.

The most important part of Agent Pay is safety. Mastercard is relying on its token system to keep transactions secure. It’s the same system used when you tap your phone to pay or store your card online. On top of that, they’ve added new steps to make sure only verified AI agents are allowed to make payments. There are also extra layers like real-time fraud detection and biometric checks on the user’s device to stop anything suspicious before it happens.

This idea of letting AI handle payments isn’t completely new, but until now, it hasn’t been very organized or secure. Mastercard is trying to change that by creating a system that’s built for how people actually use AI today. The company has already been using AI in other areas, like spotting fraud, and says this new step fits into that bigger plan.

One reason they feel confident in pushing this forward is the strong network of partners they’ve lined up. Microsoft and IBM bring the tools that many companies already rely on. Braintree and Checkout.com help with payments on the business side. Together, these companies make it easier for Agent Pay to be used in real-world situations.

For users, control is a big deal. You can decide what your AI is allowed to do. Maybe you permit it to spend up to $100 without asking you every time, but anything over that needs your approval. Merchants will also be able to see that a transaction came from an agent, not a human, which helps with tracking and reporting.

At the same time, Mastercard is keeping regulators in mind. They know there are questions about how AI should interact with financial systems, especially if it’s making decisions and spending money. So the company is adding clear rules and safety features to prevent fraud, misuse, or errors.

For now, the focus is mostly on business use. It’s easier to roll out complex tools in environments where companies already use AI for tasks like procurement. But Mastercard says consumer features will follow, and we could start seeing Agent Pay in personal digital assistants and apps soon.

There’s also competition on the horizon. Visa is working on something similar. They’ve teamed up with companies like OpenAI and Samsung, looking at how AI agents could be used in consumer electronics and digital assistants. The difference is that Mastercard’s version is built on an already-proven token system, which could make adoption faster and safer.

When you look at the bigger picture, this isn’t just about speeding up shopping. It’s about shifting how transactions happen altogether. Instead of going to a website, entering your card details, and waiting for confirmation, you might just tell an AI agent what you need and let it handle the rest. If it can do it safely and within the limits you set, that saves time and effort.

There are still a few hurdles ahead. Businesses will need to upgrade their systems to recognize and process these agent-led payments. Customers will need to feel confident that their AI assistant won’t buy the wrong thing or overspend. And regulators will need to make sure the systems are fair, transparent, and accountable.

But Mastercard says it’s ready for all of that. Their system can track agent behavior, resolve disputes, and stop fraud before it happens. The fact that they’re using existing infrastructure, like the same token technology behind mobile payments, gives them a head start.

Jorn Lambert, Mastercard’s chief product officer, said the company is moving toward a new way of handling payments by focusing on what consumers will need shortly. With the introduction of Agent Pay, Mastercard is starting to reshape how people and businesses interact with AI in everyday transactions. A key part of this includes giving merchants tools to identify which AI agents are trustworthy and which may pose risks.

He also pointed out that as this shift gains momentum, the industry needs to come together to set clear standards for how these AI-driven payments should work. One example is using protocols like the Model Context Protocol within Secure Remote Commerce, which helps create a reliable system for scaling agent-led transactions and making them more secure for everyone involved.

Mastercard recently also agreed to invest $300 million in Corpay’s cross-border business, securing roughly a 3% equity stake in a unit valued at about $10.7 billion.

As part of this deal, Corpay becomes the exclusive provider of high-value, account-to-account cross-border payments and currency-management tools for Mastercard’s banking clients. In return, Mastercard will offer its virtual card programs exclusively through Corpay to corporate customers.

About Mastercard

Mastercard Inc. is an American multinational technology company in the global payments industry, headquartered in Purchase, New York. Founded in 1966 as Interbank/Master Charge and rebranded as MasterCard in 1979 (now stylized “Mastercard”), the firm operates a real-time transaction processing network that connects issuers, merchants, and consumers across more than 210 countries and territories. Its suite of products—including credit, debit, and prepaid cards, as well as value-added services like fraud prevention, tokenization, and data analytics—facilitates the movement of funds and information, enabling more than 2.5 billion cards in circulation and trillions of dollars in annual purchase volume.

A publicly traded company on the New York Stock Exchange under the ticker “MA,” Mastercard delivered net revenues of $28.2 billion in fiscal year 2024, up 12% year over year, with net income rising 15% to $12.9 billion. Building on this financial momentum, the company continues to broaden its technological edge through strategic acquisitions, such as the cybersecurity specialist Recorded Future, to bolster its fraud prevention and intelligence offerings, while investing heavily in AI and tokenization to drive the next generation of digital payment solutions. Beyond its core business, Mastercard advances social impact via the Mastercard Foundation, which holds assets of approximately $47 billion and supports education and economic development initiatives in nearly 50 countries.

Conclusion

Agent Pay marks a clear step toward practical, secure AI-driven payments. By combining trusted token technology with new controls for machine-led transactions, Mastercard is building a system that fits how people and businesses are starting to use AI in real life.

With strong partners and a focus on safety, the company is laying the groundwork for AI agents to take on more responsibility in everyday commerce, while keeping users in control.