FICO has published a study highlighting that one in three organizations report high or very high false positive rates in AI-based protection tools.
This matters because false positives are not just an operational nuisance. When legitimate transactions are flagged as suspicious, customers face delays, extra verification steps, or denied services. That friction erodes trust, harms the customer experience, and increases customer churn, especially in fast-growing digital markets. As a positive note, the FICO study suggests that the region’s relatively solid performance in the EMEA region (up to 20 percentage points lower than other regions in false positives) is driven by regulatory maturity and optimized systems, which have proven capable of significantly reducing this problem.
AI, a Facilitator of Large-Scale Fraud
Fraud attempts have risen significantly, according to the FICO study, with 26% of organizations reporting an increase in fraud of more than 51% over the last two years. AI has lowered the barrier to entry for fraudsters, fueling the creation of synthetic identities, account takeovers via deepfakes, systematic and automated bot attacks powered by AI, and large-scale social engineering attacks aimed at deceiving potential victims, previously possible only for sophisticated criminal networks.
EMEA data show more moderate growth in fraud globally (48% report increases between 11% and 25%) and a smaller share of reports with losses above 51% (12%), suggesting that stronger regulatory frameworks and greater detection maturity are having a measurable effect.
And while 99% of organizations acknowledge the need for a unified fraud strategy, only 28% have fully deployed AI-driven fraud detection at scale. The biggest hurdle isn’t the technology—it’s integration. 47% of fraud leaders surveyed say the challenge is effectively integrating AI into their existing fraud-control frameworks.
According to the FICO report, only 34% have fully deployed any of the three main AI use cases analyzed: generative AI for customer support, generative AI for case-management assistance, and AI/ML-based detection models. Most financial institutions remain in pilot stages or limited production deployments.
Fragmented or siloed fraud tools leave gaps even when sophisticated individual models are used. The organizations best positioned to succeed are those that build unified decision-making frameworks that orchestrate internal models, external provider intelligence, and real-time consortium data across channels, products, and customer journeys. In fact, nearly half of organizations (47%) cite the difficulty of effectively integrating AI as their top challenge, ahead of the growing complexity of fraud (40%), regulatory constraints (35%), data governance issues (34%), and legacy-system limitations (32%).
