KYC: Bypass age verification using generative video models (www.synacktiv.com)

Recent developments have highlighted the vulnerabilities in the Know Your Customer (KYC) processes, particularly regarding age verification. With the increasing deployment of these systems across online platforms for compliance and user trust, generative video AI models have emerged as tools that can effectively bypass age verification checks. A study demonstrated a successful circumvention of age verification on a website utilizing AWS Rekognition, leveraging techniques that inject pre-recorded videos or AI-generated content during the identification process. This raises significant concerns about the robustness of current KYC methodologies, which are designed to ensure user authenticity via "proof of life" tests. The significance of these findings for the AI and machine learning community is immense, signaling that existing verification frameworks must evolve to counteract emerging AI capabilities. The PVID framework in France is attempting to bolster defenses by mandating stringent measures, such as sensor attestation and motion data analysis, to distinguish between real-time images and manipulated content. However, the study illustrates the potential for attackers to exploit weaknesses in these systems using accessible technology, urging a re-evaluation of KYC protocols to strengthen digital identity verification and improve security against identity fraud while maintaining user privacy and consent.