Effectiveness and limitations of biometric systems in identity verification: A systematic review

Abstract

Increasing digitization and automation demand advanced and secure methods for access control. This study synthesized artificial intelligence (AI) tools applied in this field, through a literature review in databases such as Scopus, SciELO and IEEE Xplore, using PRISMA and VOSviewer. The bibliometric analysis identified China, India, the United States and South Korea as leaders in research, highlighting terms such as machine learning, deep learning, cryptography and biometrics, along with emerging technologies such as blockchain and IoT. Machine learning and deep learning stood out as predominant techniques, while blockchain brought transparency in the management of sensitive data. However, challenges such as high costs, reliance on big data and privacy concerns limit its implementation. It is recommended to explore hybrid methods, optimize AI models and reduce data dependency to improve security and adoption of these technologies.