What We Feel the New Gartner® Hype Cycle™ Reports Signal About Fraud Prevention Trends

Banks are surrounded by vendors telling them AI will transform fraud prevention. What they’re short on is a reliable way to tell which of those claims are already true and which are still aspirational.

According to us, that’s the value of the Gartner Hype Cycle research. It maps where a technology actually sits, independently of where vendors say it does. In June 2026, Gartner published four new reports covering AI in banking: the Hype Cycle for Artificial Intelligence in Banking, 2026,¹ Hype Cycle for Banking Customer Experience, 2026,² Hype Cycle for Data and Analytics in Banking, 2026,³ and  Hype Cycle for AI-Driven Trade Finance Transformation in Banking, 2026.⁴

The banking-wide report opens with a striking observation: “AI adoption in banking is surging, yet only 38% of organizations see financial gains.“¹

Feedzai is thrilled to be recognized in these four reports. These three stand out in particular: Hype Cycle for Artificial Intelligence in Banking, 2026,¹ Hype Cycle for Banking Customer Experience, 2026,² and Hype Cycle for Data and Analytics in Banking, 2026.³ But in our view, the real story here isn’t about vendors, it’s what the maturity curve says about where banks should be investing now, where they should be watching closely, and where they should be more skeptical of the pitch in front of them.

Machine Learning Has Already Crossed Into the Plateau

The Hype Cycle for Artificial Intelligence in Banking, 2026, places machine learning in banking in the “Entering the Plateau” phase, with a “Transformational” benefit rating and market penetration of “more than 50% of target audience.” In our view, that’s as clear a signal as you’ll find that ML-driven fraud detection, credit decisioning, and AML monitoring are not emerging bets anymore, they’re established practice. 

But we don’t think “mature” means “solved.” One of the obstacles the report states is “black-box models reduce business, regulator, and customer trust, and limit clarity for declined transactions.”1 That’s the same problem regulators keep raising, and it’s a governance challenge as much as a technology one. 

In our view, the technology has matured faster than the model risk management and explainability practices sitting around it. For risk and compliance leaders, the question is not whether you should be adopting ML (that decision is years behind you), but whether your model risk management and explainability practices have kept pace.

Simulation Is the Layer Most Banks Haven’t Built Yet

The placement that stands out to us, because it appears in all new reports, is AI simulation in banking. Gartner rates it as “High” benefit at “Adolescent” maturity, with the priority matrix in all reports placing it “5 to 10 years” from mainstream adoption.¹ AI simulation is becoming especially important in the Agentic era, with more diverse use cases rapidly being created.

Simulation solves a problem that’s specific and stubborn in financial crime: the data you most need is the data you have the least of. Fraud and money laundering are rare-event, imbalanced problems by nature, and historical data alone can’t show you the next attack pattern. 

We believe the drivers listed in the reports for this category map closely to where fraud teams actually get stuck – generating realistic synthetic data where real data is scarce, stress-testing models against scenarios that haven’t happened yet, and proving to regulators that a model holds up under conditions beyond what it was trained on. Feedzai is named in Predictive Analytics in Banking in the Hype Cycle for Data and Analytics in Banking, 2026,³ and AI Simulation in Banking in the Hype Cycle for Artificial Intelligence in Banking, 2026.¹

Simulation and synthetic data is not only a fundamental feature of the Feedzai Platform, but also freely available as part of our open source code base available to everyone. In our view, this matters because the technology already protecting banks today and the technology that will matter most over the next five-plus years aren’t the same thing. Banks investing only in the first are optimizing for the fraud patterns of yesterday.

What’s Overhyped Right Now and Why That’s Useful to Know

In our opinion, one of the most useful elements of these reports is their clarity on what’s currently overpromising.

In the Hype Cycle for AI-Driven Trade Finance Transformation in Banking, 2026, technologies such as agentic AI in banking and agentic analytics — both sitting atop the peak — are central to this transformation, empowering banks to streamline operations and optimize transaction execution.”⁴

The same report notes that notably, AI sales assistants have moved into the Trough of Disillusionment, signaling a period of reassessment following initial excitement, whereas AI agents in payments are climbing toward the Peak of Inflated Expectations, driven by growing enthusiasm and high expectations for their transformative potential. Gartner observes that “Generative AI in Banking has moved into the Trough of Disillusionment.“¹ This position reflects a more grounded understanding of its limits and the effort required to scale it.

We believe none of this means these technologies won’t eventually deliver and the report tends to agree, with several carrying long-term transformational ratings in all four reports. But in our view, a vendor pitching production-ready autonomous payment agents today is selling further out on the curve than the market actually is. 

That gap between what’s deliverable now and what’s still being proven is precisely what a hype cycle is designed to surface, and it’s a useful reality check before any roadmap conversation built more on vendor promises than current capability.

The Pattern Worth Taking Into 2027 Planning

We think these  reports, read together, tell a consistent story. The technology actually protecting institutions today may not be glamorous, but it is mature and, increasingly, under pressure to be explainable. 

The technology worth watching next is simulation and the ability to build the resilience layer that lets fraud teams prepare for what historical data can’t show them. 

That’s a more useful lens for a 2027 AI roadmap than any vendor’s name on any list.

GARTNER and Hype Cycle are a trademark of Gartner, Inc. and/or its affiliates.

Gartner does not endorse any vendor, product or service depicted in its research publications, and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner’s research and advisory organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.