Hear from the VP/GM of Payments at Google and the Chief Digital Officer at Mastercard on what the future of agentic payment holds — and what needs to happen first.
The age of AI is advancing faster than anyone imagined. Not long ago, it was all about chatbots or digital assistants that answered questions, explained concepts, and mimicked conversation. Now, those systems have transformed into agentic tools that can reason, plan, and act.
They browse the web, book appointments, summarize meetings, and soon, may even handle tasks we consider deeply personal.
This so-called “agentic shift” is spawning a wave of tools like Comet, Dia, and Jace AI. Most recently, OpenAI made its move in the category by introducing Atlas – a browser that can explore the internet and even step in to complete tasks for you autonomously.
But as these AI agents begin to take real-world actions, tougher questions emerge. The biggest one: What happens when an agent needs to do something only a human could — like making a payment?
For now, most systems stop short of that threshold, prompting users to intervene and complete the payment. However, as agents become more capable and machines increasingly navigate the web rather than people, that dependency on a human for authentication would fall, too.
To prepare for that future, technology and payments giants have already started building the rails for secure, verifiable agentic transactions. Their goal isn’t to make AI agents spend freely, but to ensure that when they do handle your money, every click, credential, and charge can be traced and trusted.
The foundation for trust
Leading the development of this new-age payments experience are Google and Mastercard, which are both defining standards and security rails for how agents pay.
Google has developed the Agent Payments Protocol (AP2), an open framework that enables agents to securely make purchases for users, with the backing of over 60 financial and tech giants. Stavan Parikh, the VP/GM of Payments at Google, described it as a “common language” that helps agents smoothly work with merchants to transact.
At its core, AP2 works by using Intent and Cart Mandates. These are cryptographically signed digital contracts that record a user’s authorization to search for a product and pay for it before eventually serving as tamper-proof, verifiable evidence of their instructions. The mandates cover both real-time purchases and scheduled ones, where you define specific rules of engagement (like buying only when an item goes on sale) for the agent to execute.
“AP2 enables entirely new commercial experiences that are more complex and coordinated,” Parikh told Future Nexus, adding that “in the future, a user could simply instruct their AI agent to book a round-trip flight and a hotel for a specific weekend with a total budget of $1,500. The agent could then find a combination that fits and securely execute both bookings simultaneously.”
Parikh emphasized that the problem of trust and accountability is the biggest one holding agentic commerce back. But with AP2, its massive roster of partners and secure, cryptographically signed data objects, Google is solving for the trust question, ensuring users retain authority with transparent control over their agents’ financial activities. The framework also ensures that the roles of the shopping agent and the credentials provider (e.g., a digital wallet) remain separate to prevent unnecessary access to payment information.
“By providing a single, open standard, we eliminate the need for costly, bespoke integrations and create a clear path for merchants of all sizes to use agentic commerce. We are making the specification and reference implementations publicly available to accelerate adoption,” he added.
As AP2 continues to evolve, payment network Mastercard is adding support for it while also building out its own stack with Agent Pay, which registers agents and then governs them using Mastercard’s own tokenization technology that is already securing billions of mobile and online payments.
“Agentic Tokens enable agentic payments via conversational interfaces and at millions of merchants, giving everyone in the value chain — consumers, issuers, and merchants — clear visibility into agent-facilitated transactions and delivering greater transparency,” Pablo Fourez, Chief Digital Officer at Mastercard, told Future Nexus.
These dynamic, verifiable tokens are being developed in alignment with protocols like AP2. Plus, Mastercard is working with the FIDO Alliance and its Payments Working Group to establish verifiable credential standards that can confirm key payment details like amount, merchant, and product, ensuring every AI-initiated transaction remains “shopper-approved.”
The payments giant has also partnered with Microsoft and IBM to integrate their AI technologies and systems with Agent Pay.
“Payments are quickly becoming essential to how AI platforms interact with the world. For Mastercard and our partners, that means evolving from enabling digital transactions to enabling digital decision-making. Financial players will need to provide not only the infrastructure for money movement but also the frameworks, credentials, and APIs that make AI-powered transactions secure and interoperable. With our end-to-end solutions, we’re helping banks, developers, and merchants integrate payments safely into AI-driven experiences. The goal is not only to adapt to this technology but to shape it responsibly,” he added.
Notably, Google and Mastercard aren’t alone in this.
OpenAI, the leader in the AI category, is also taking a step toward agentic commerce, enabling instant checkout via ChatGPT, though it’s still authenticated manually by users right now instead of the AI handling everything end-to-end. Visa, the leader in payments, has also announced Intelligent Commerce, which uses AI-ready cards with tokenized credentials and user-set preferences (such as spending limits and conditions) to enable AI agents to find and purchase items without exposing card data.
All these companies are laying the foundation for secure, permissioned AI transactions, where agents can act autonomously and pay for you, but with complete transparency and while staying within the rules defined, verified, and enforced by the user.
Who bears the liability when AI goes wrong?
Despite the measures to ensure trust, security, and accountability, it’s fair to assume that agentic transactions will go wrong, at least occasionally. Maybe AI would misconfigure the user’s intent and buy an incorrect item, it could calculate an incorrect bill amount and pay only half of the total, or maybe a new kind of fraud emerges with agentic payments. Those possibilities are yet to be explored.
In all these (and potentially other) scenarios, traceability will come in handy, and so will the liability rules — who will be held responsible when something goes wrong.
According to Google’s Parikh, AP2 adds accountability, but it won’t rewrite the rules of liability, which continue to be governed by payment networks and banks.
“The protocol creates an unalterable audit trail and clear accountability by anchoring every transaction to cryptographically signed user mandates and verifiable credentials. This allows merchants, banks, and consumers to transact with confidence. It establishes a clear visibility and accountability framework that gives banks, merchants, and payment networks the confidence to participate,” he said.
Fourez also acknowledged that as AI agents begin to transact on behalf of consumers, new risks will emerge, such as compromised credentials and unauthorized agent behavior. However, instead of diving deep into the liability details, he emphasized embedding security at every interaction of the agentic payment flow and keeping users aware of how agents will pay for them.
“Making agentic payments truly secure will also require collaboration across the industry, with platforms, merchants, payment networks, and standards bodies establishing consistent protocols so users can transact confidently. We’re also leveraging our deep services capabilities — including our threat intelligence, monitoring, and fraud prevention solutions — to defend against ongoing and emerging payment threats,” he said.
For people, he said, agentic transactions won’t be about learning new technologies — they are already used to seamless checkout experiences, given the rise of digital payments. The learning curve, he said, would be more about knowing when and how AI could act on their behalf and how they can stay in control every step of the way.
The dark alley ahead
Even though AI agents can only act where the humans in the loop allow them to, the more capable they become, the less visible our permissions will be. The future of payments will soon be the story of defining where exactly that loop begins and ends.
There’s also a darker alley running parallel to this progress: the AI-driven payments space is filled with many unknowns about how AI will behave when it’s truly in the driver’s seat with our money. That’s what makes the work of companies like Google, Mastercard, Visa, and OpenAI so critical.
Before these pilot projects and research efforts become full-scale systems, the industry has to get the guardrails right. Once AI starts paying on our behalf, righting the wrongs will prove much more difficult.
Hear from the VP/GM of Payments at Google and the Chief Digital Officer at Mastercard on what the future of agentic payment holds — and what needs to happen first. The age of AI is advancing faster than anyone imagined. Not long ago, it was all about chatbots or digital assistants that answered questions, explained concepts, and mimicked conversation. Now, those systems have transformed into agentic tools that can reason, plan, and act. They browse the web, book appointments, summarize AI, Home, News, Popular, agentic commerce, AI financial agents, AI payments, conversational commerce, digital payment security, FIDO Alliance, Google AP2, Mastercard Agent Pay, OpenAI Atlas, Pablo Fourez, Stavan Parikh, tokenized payments, verifiable credentials, Visa Intelligent Commerce
This articles is written by : Nermeen Nabil Khear Abdelmalak
All rights reserved to : USAGOLDMIES . www.usagoldmines.com
You can Enjoy surfing our website categories and read more content in many fields you may like .
Why USAGoldMines ?
USAGoldMines is a comprehensive website offering the latest in financial, crypto, and technical news. With specialized sections for each category, it provides readers with up-to-date market insights, investment trends, and technological advancements, making it a valuable resource for investors and enthusiasts in the fast-paced financial world.