The rapid integration of algorithmic facial recognition technologies (FRT) into the activities of the administrative police generates a deep institutional imbalance. The formation of a total video surveillance infrastructure poses a direct threat to citizens' constitutional rights to privacy and freedom of movement, exacerbated by the lack of clear specific legislation and the risks of machine bias. A comprehensive conceptual analysis of existing legal models for the use of biometric technologies by the police to develop scientifically grounded recommendations for achieving a balance between public safety and the informational autonomy of the individual. The research applies a comprehensive approach, including systemic, formal-dogmatic, and comparative-legal (EU, USA, UK, RK) methods. Special emphasis is placed on the analysis of the ECtHR case law (Glukhin v. Russia) and the relevant empirical statistics of the Ministry of Internal Affairs of the Republic of Kazakhstan for 2024–2025. The high forensic effectiveness of FRT is recognized. However, it has been established that in Kazakhstan, the use of recognition systems is primarily regulated by subordinate acts, which provides law enforcement agencies with excessive discretion. Indiscriminate biometric scanning of the population de facto erases the presumption of innocence. The critical need to reform national legislation by implementing international risk-oriented standards (EU AI Act) and ECtHR principles (extrema ratio) is substantiated. Specific steps are proposed: mandatory judicial authorization of algorithmic searches, establishment of an exhaustive list of crimes for FRT use, and the creation of an independent AI audit institution.