Biometric authentication has changed how establishments verify identity, secure access and also have become indispensable components of modern security infrastructures. Nevertheless, they are still extremely susceptible to presentation attacks such as spoofing through printed photographs, replayed videos, and synthetic facial masks. To mitigate these vulnerabilities, this study suggests an AI-based multimodal face Liveness framework that leverages both RGB and Infrared (IR) imaging for enhanced and secure biometric authentication. The integration of RGB and IR modalities enables the system to capture complementary facial information. RGB images provide rich texture and color details, while IR imaging reveals physiological and sub-surface skin characteristics that are difficult to replicate in spoofing attempts. The proposed method employs deep learning-based feature extraction using Convolutional Neural Networks (CNNs) to learn discriminative spatial representations from both modalities. A fusion mechanism is applied to combine RGB and IR feature representations, enabling more robust Liveness classification. The model is trained on a multimodal dataset containing genuine and spoof facial samples collected under varying environmental conditions. Experimental results indicate that the proposed multimodal approach outperforms single-modality systems in terms of accuracy, precision, and robustness against diverse attack scenarios. Furthermore, the incorporation of infrared imaging improves system performance in low-light and illumination-variant environments, making it suitable for real world deployment in security critical applications such as mobile authentication, border control, and financial systems. Generally, the findings demonstrate that multimodal fusion of RGB and IR data, combined with AI-driven learning techniques, delivers a reliable and scalable solution for next-generation biometric Liveness systems.