Oz Forensics has achieved ISO/IEC 30107-3 Level 3 certification for its OzLiveness face presentation attack detection SDKs, the company has announced.
BixeLab, an NVLAP-accredited biometric testing laboratory based in Canberra, Australia, conducted the evaluation between February and March of this year. The lab tested OzLiveness SDK for iOS (version 9.0.0) and Android (version 9.0.1), alongside the OzLiveness Web SDK (version 1.9.2). Testing was performed on Android 16 and iOS 26 mobile devices, with browser-based evaluation carried out on Safari version 26.2 and Chrome version 145.0.7632.76.
Level 3 represents a higher tier of presentation attack detection sophistication, involving carefully crafted attack instruments designed for specific test subjects. BixeLab has previously conducted PAD evaluations for other identity vendors, including FaceTec, which completed Level 1 through 5 testing in late 2025. BixeLab conducted 100 Level C presentation attack tests across four attack types using five consenting test subjects for the Oz Forensics evaluation. An additional 50 bona fide transactions were completed across 10 subjects to measure false rejection rates.
Both metrics returned zero classification errors. The OzLiveness systems achieved a zero percent Attack Presentation Classification Error Rate (APCER) and a zero percent Bona Fide Presentation Classification Error Rate (BPCER) across the test set.
The certification confirms that OzLiveness meets the ISO/IEC 30107-3 standard for PAD testing at Level 3 sophistication, the same standard used by governments and enterprises evaluating biometric onboarding systems for security-sensitive applications.
Oz Forensics completed a SOC 2 Type 2 examination for its biometric services in November 2025, validating its data security and operational controls for identity verification platforms.
OzLiveness is used by financial institutions, government agencies, and fintech companies to verify that a face presented during identity onboarding is a live human and not a photo, video, or mask attack.
Sources: BixeLab
–
By Ji-seo Kim
