The National Institute of Standards and Technology updated its Face Analysis Technology Evaluation (FATE) Age Estimation and Verification (AEV) results as of February 13, 2026, adding performance data for two newly submitted algorithms in its ongoing assessment of software that analyzes facial images to estimate age.
In the updated report, the two new algorithm entries are ROC’s roc-002 (submitted February 6, 2026) and Shufti’s shufti-000 (submitted January 26, 2026). The additions extend NIST’s continuous evaluation model, under which developers may submit updated algorithms on a recurring basis, but no more frequently than every four months.
From one-off report to continuous testing
NIST released its first age estimation evaluation in a decade in May 2024, positioning the work as the start of a longer-term effort to track progress in age estimation and verification technology under repeatable test conditions. The AEV track sits within the agency’s broader face technology evaluation program, which was reorganized into two branches: FRTE for face recognition algorithms and FATE for face analysis technologies such as age estimation.
The AEV evaluation measures accuracy primarily using Mean Absolute Error (MAE), summarized as an average of 26 MAE values across two sexes and 13 age ranges spanning 18 to 30. NIST also reports computational efficiency metrics, including processing time and model size, to help characterize operational tradeoffs across submissions.
Vendors continue to iterate as age assurance demand grows
Submissions to the AEV track have included a mix of biometric and identity technology vendors and research groups, including Innovatrics, Yoti, Regula, Ondato, Cognitec, CVUT, CyberLink, Dermalog, Incode, Neurotechnology, ROC, Unissey, and others. The continuous format is designed to capture iterative improvements over time, with vendors resubmitting updated models as methods evolve and performance is optimized for different operating conditions and populations.
That momentum mirrors rising regulatory and industry interest in age assurance. Facial age estimation is increasingly being positioned as a pathway to compliance for age-gated services and minor protection requirements across online services. As deployments expand, scrutiny has also increased around performance consistency across demographic groups and the risk of disparate error rates across age bands, sex, and other characteristics.
In prior updates, Regula ranked among the stronger performers in NIST testing in policy-relevant scenarios such as Challenge-25 style thresholds and child online safety contexts. Ondato also appeared in the NIST evaluation, with NIST reporting results intended to support comparative assessment rather than certification or approval.
As NIST continues to publish updated results, the AEV track is becoming a recurring reference point for organizations evaluating age estimation tools, particularly as accuracy requirements, fairness expectations, and operational constraints tighten in real-world deployments.
Sources: NIST FATE AEV, NIST AEV Report (NISTIR 8525), NIST Technical Report
–
By Ji-seo Kim
