Case Study: Dutch Digital Bank reduces deepfake ID fraud with instinctools’ computer vision layer

Dutch Digital Bank cuts false positives by 60% with instinctools

a Dutch digital bank faced a growing threat from AI-generated deepfake identity documents used to bypass its existing third-party SaaS ID verification (IDV) platform. This gap posed significant financial and reputational risks as the platform was slow to adapt to new fraud patterns and generated too many false positives for legitimate users. The bank partnered with *instinctools to address this challenge.

*instinctools implemented a custom computer vision-based deepfake detection system to run in parallel with the bank's existing IDV platform. This YOLO-based pipeline was trained on a robust, fraud-realistic dataset and integrated seamlessly into the KYC flow. The solution led to a 30% increase in detected synthetic document fraud, a 60% reduction in false positives, and a 40% decrease in the manual review workload.