By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions.
Nowhere is its influence more immediate and more invisible to the average consumer than in the payment systems that underpin everyday commerce.
From streamlining back-office operations to reinventing fraud prevention, AI is shifting from a peripheral tool to the backbone of global payments infrastructure.
To understand how this transformation is playing out, PYMNTS interviewed two executives who are helping to chart the industry’s course: Judith McGuire, senior vice president of global products at Discover Network, and Ian Hillis, senior vice president of growth at Worldpay.
“I would say there’s probably three areas where we’ve seen some pretty important benefits of AI,” McGuire said. “First in customer service … helping with scripting, prompting and creating more information for the customer service agent so that they can have a much more impactful conversation with the customer.”
“Another area where we’ve seen a lot of benefit from AI is in back-office tasks, freeing up our folks from repetitive tasks and enabling them to do more important work,” she added.
“AI is helping us unlock the power of data that we can then use to monitor the network, to reduce fraud, to help make decisions even faster,” McGuire said, adding that human oversight, testing and AI deployment guardrails are “absolutely essential.”
“If you think briefly about the software platforms, software platforms have been the early adopters of AI, and that can manifest through engineering, efficacy, support within the platform, as well as [a] host of other use cases specific to payments,” Hillis said. “These platforms typically aren’t looking to reinvent payments. They want to grow fast … and AI helps abstract away the complexity that can be associated with payments while including the advantages mentioned earlier.”
The most pressing challenge is security, McGuire and Hillis said. Payments fraud has always been a game of cat and mouse, but generative AI tools are giving fraudsters unprecedented capabilities.
“Unfortunately, the same advantages AI provides of speed, improved actions, lower cost and better fidelity apply to the fraudsters as well,” Hillis said. “Anyone can hop onto any free-to-use gen AI tool, and pretty quickly, you can have a realistic avatar and supporting credentials. Five years ago, that took significant resources and time. Today, we have to be the ones to stay ahead of that.”
McGuire described how Discover Network is adapting.
“Certainly, AI is essential to contributing to security of the network,” she said. “One way we use it is to monitor the network [and] make sure that the folks transacting are the ones that we want.”
AI models, trained on vast transaction datasets, are capable of detecting anomalies in real time. They evaluate hundreds of variables, such as location, device fingerprint, transaction history and subtle behavioral cues to assess the probability of fraud in milliseconds. For global platforms processing millions of daily payments, this is the difference between scalability and stagnation.
“Fraud defense lives and dies on data,” Hillis said. “We’ve seen some crazy stuff, and we learn quickly from it. For nearly a decade, we’ve been embedding AI across core payments infrastructure, optimizing and preventing fraud. These models are trained in 55 billion transactions annually, 135 currencies, 174 countries.”
Both leaders agreed that data management is the linchpin of AI success.
“The AI revolution will fundamentally change how companies manage data,” McGuire said. “They’re going to have to make data accessible, understandable and scalable in a much different way than they have previously … managing data is going to be the core to facilitating this AI revolution.”
For Hillis, the metaphor is culinary.
“AI without data is like a recipe without ingredients,” he said. “AI with mediocre data is a recipe that’s going to be fed to the dog under the table. While modern AI has reduced the need for perfect data, the better the ingredients, the better the dish.”
At the same time, for all the progress, McGuire and Hillis admitted the payments industry’s use of AI is only beginning.
“It’s going to come down to making sure we’ve got thoughtful policy and delivery approaches in place as an ecosystem,” McGuire said, pointing to the need for ecosystem-wide collaboration. “Making sure we’re all working together to create policies and delivery mechanisms that our end customers can trust, that’s going to be key.”
“AI in payments isn’t perfect yet, and every new breakthrough is going to carry unresolved friction,” Hillis said. “Training and tuning are critical. And transparency gaps remain. Black box AI can backfire if merchants can’t explain outcomes.”
If the rise of machine learning in fraud prevention was evolutionary, the maturation of the agentic capabilities of AI systems, or their capacity for autonomous action, could be revolutionary.
“Agent-led commerce is about the closest thing I’ve seen in payments to sci-fi magic, and it really has the potential to flip the world of payments on its head,” Hillis said. “Your trust isn’t in the cart anymore. It’s in the underlying infrastructure.”
“From a merchant lens … we expect to see supercharged conversion rates, minimization of cart abandonment,” he added. “Agents aren’t going to forget to buy something. They’re also going to optimize price and supplier match. So, merchants are going to get access to new customer segments that potentially lower [customer acquisition cost (CAC)].”
McGuire said she also sees the conversation around agentic commerce turning toward fundamentals.
“It’s going to be around trust,” she said. “We need to create the framework for this whole new environment where you can have agents impersonating cardholders, merchants, acquirers. How do we build a truly trustworthy environment in this new world?”
“Creating that trust is going to be key to getting consumers engaged on a wide scale,” she added.
