Paravision Achieves Perfect Demographic Precision in Independent Age Estimation Bias Evaluation by ACCS

San Francisco – April 10, 2025 – Paravision, a leader in trusted Identity AI, today announced the results of a new independent bias evaluation conducted by the Age Check Certification Scheme (ACCS) on its Age Estimation technology. The findings confirm that Paravision Age Estimation delivered 100% precision across all tested demographic groups, setting a new benchmark for fairness and reliability in AI-based age assurance and underscoring its readiness for age-restricted digital services such as alcohol delivery, online gaming, and e-commerce platforms.

Paravision recently announced it received ACCS’s highest-level certification—Level 3 “Highly Effective Compliance”—with a 100% precision score on Challenge 25, the industry benchmark for underage identification. The ACCS Bias Evaluation is an optional extension to ACCS’s Level 3 certification process and tested Paravision Age Estimation accuracy, precision, and bias across gender and skin tone groups, yielding exceptional results across lighting conditions and a testing base of hundreds of individuals aged 17-20.

In summary, the reliability of the system’s precision, supported by a narrow confidence interval, underscores its effectiveness and trustworthiness in real-world applications, where accurate and consistent age estimation is essential. This high level of reliability assures users and decision-makers that the system can be confidently relied upon to perform as expected, with minimal risk of significant error.
ACCS Bias Evaluation Report, April 2025

Precision by Demographic Group

In addition to Paravision achieving 100% precision across all tested demographic groups, ACCS also evaluated Mean Absolute Error (MAE) rates within these groups. Each group’s individual MAE was reported, and Paravision performed well below the maximum passing threshold of 2 years for every group, as detailed in the table below.

This report confirms that our Age Estimation technology not only achieves industry-leading accuracy but does so with highly consistent performance across gender and skin tone, and in varied lighting. Fairness and reliability are critical in AI systems – especially those used for identity and compliance – and we’re proud to embrace their importance.
Joey Pritikin, Chief Product Officer at Paravision

Paravision Age Estimation is available in SDK and Docker formats and enables partner solutions for compliance with privacy, security, and regulatory requirements. For access to the full ACCS report or integration inquiries, contact us at [email protected].