Trulioo upgrades its proprietary models for improved verification rates

The company identified that fraud techniques such as ID alteration and deepfakes account for 16% of all cyberattacks. In response to this challenge, Trulioo built AI and machine learning models to fend off sophisticated fraud attempts while ensuring accuracy and minimising friction for its users.

The company is committed to continuing to refine these solutions while remaining compliant with the regulation requirements and laws of the industry. Trulioo built its proprietary machine learning models to respond faster to emerging threats and its customers’ evolving needs.

Faster and smarter verification with AI features

The latest updates to Trulioo’s proprietary AI and machine learning models in the last six months include a 20% increase in verification rates, helping businesses onboard legitimate customers. The models also present a 60% reduction in job processing times, allowing for faster identity verification, and a 20% increase in auto-approval rates. This helps to minimise manual reviews and reduce friction for users while decreasing costs for organisations.

Trulioo’s AI and machine learning features capture multiple images per second to increase fraud detection in real-time and ensure accuracy in document verification and biometrics automatically. Alongside autocapture, they also utilise face matching by leveraging deep learning models and convolutional neural networks to extract and analyse unique facial features, calculating a similarity score for precise identity verification.

Other features included are intelligent document and liveness authentication and fraud detection. The AI models detect and analyse documents and selfies for signs of forgery, deepfakes, and digital manipulation, recognising anomalies in security features such as holograms, watermarks, and microprinting. They also apply machine learning to verify ID documents across the globe by detecting security features, extracting key information and ensuring the authenticity and liveness of the document.