Case Study: Onfido Achieves 10x AI Fraud Detection Improvement with Appen

How Onfido's labeling tool enhanced machine learning

Onfido, a global leader in AI identity verification, needed a secure and scalable way to improve fraud detection across sensitive biometric, image, video, and document data. Working with Appen, they sought an on-premise data labeling solution to protect PII, meet privacy requirements, and support the training of more accurate machine learning models for identity verification.

Appen provided Onfido with a custom on-premise AI Data Platform and flexible annotation tools tailored to their workflows, along with ongoing collaboration and API-based human-in-the-loop retraining. This enabled Onfido to label large volumes of fraud data securely and improve the performance of its fraud detection models by 10x, helping strengthen the Real Identity Platform and better detect sophisticated fraud tactics.

Onfido

François Jehl

Director of Data and Analytics