Eika Group

Eika Group, a consortium of 46 community and regional banks across Norway, faced a convergence of fraud pressures that exposed the limitations of legacy rule-based detection. Phishing attacks were rising in both frequency and sophistication, with elderly customers disproportionately targeted—a pattern common in community banking where trust-based relationships can be exploited. Fraud schemes were evolving faster than manual rule updates could accommodate, allowing increasingly complex patterns to slip through existing filters. Meanwhile, analysts across the 46 member banks were absorbing growing volumes of manual transaction reviews, straining operational capacity. The cumulative effect was measurable financial loss and deteriorating customer protection across the consortium's shared infrastructure.

Eika deployed Featurespace's ARIC Risk Hub—an adaptive AI platform recently acquired by Visa—to replace static rule-based detection with continuous behavioural modelling. The system applies anomaly detection and pattern recognition by building individual behavioural profiles for each customer, establishing a dynamic baseline of normal activity against which every transaction is evaluated in real time. Every payment initiated through Eika's online or mobile banking channels is assessed before completion; suspicious transactions are automatically flagged or blocked. Critically, the platform operates as an open ecosystem compatible with all payment rails—not just Visa's network—giving the consortium unified fraud coverage across its entire payment infrastructure. Jon Hagen, Chief Architect at Eika, noted that as fraud patterns evolve, adjusting detection rules is straightforward within the system's framework, reducing the burden on internal teams to keep pace manually.

Between 2023 and 2024, Eika achieved a 90% reduction in losses from phishing attacks—the primary fraud vector targeting its customer base. Beyond that headline figure, the adaptive behavioural modelling delivered meaningful secondary outcomes:

• Fewer false positives: legitimate transactions were less frequently interrupted, reducing friction for customers and avoiding unnecessary account holds
• Reduced operational load: manual transaction review volume decreased across all 46 member banks, freeing analyst capacity
• Improved protection for vulnerable segments: elderly customers—previously the most heavily targeted group—benefited from detection logic that adapts to individual behaviour rather than relying on population-level rules

The results demonstrate that adaptive AI can address both financial and operational dimensions of fraud simultaneously.

• Adaptive AI outpaces static rules: behavioural profiling that updates continuously allows detection to keep pace with evolving fraud tactics without requiring constant manual rule maintenance.
• Consortium deployments require open-ecosystem platforms: shared infrastructure across multiple member institutions demands fraud coverage that spans all payment rails, not a single network.
• Protecting vulnerable segments drives measurable ROI: targeting fraud patterns that disproportionately harm elderly customers creates a clear, quantifiable business case for AI investment.
• False positive reduction is as important as fraud catch rate: minimising friction for legitimate customers is a key success metric alongside loss prevention.
• Vendor acquisition context matters: Featurespace's integration into Visa's value-added services portfolio signals broader institutional backing and long-term platform continuity.