This paper proposes an innovative approach to deepfake video detection by integrating features derived from ant colony optimization–particle swarm optimization (ACO-PSO) and deep learning techniques. The method extracts ACO-PSO features from the spatial and temporal characteristics of video frames, capturing subtle patterns indicative of deepfake manipulation. These features are then used to train a deep learning classifier to distinguish between authentic and deepfake videos. Extensive experiments on benchmark datasets demonstrate the method's superior performance in terms of detection accuracy, robustness to manipulation techniques, and generalization to unseen data. The approach achieves an accuracy of 98.91% and an F1 score of 99.12%, indicating its effectiveness in deepfake detection. The integration of ACO-PSO features and deep learning enhances the precision and resilience of deepfake detection, addressing challenges in facial forgery detection and contributing to the safeguarding of digital media integrity. The method is computationally efficient and suitable for real-time applications. However, it faces limitations such as dataset size, computational complexity, robustness to new manipulation techniques, and ethical considerations. Potential solutions include expanding the dataset, optimizing the model, developing adaptability mechanisms, and establishing ethical frameworks. The study concludes that the proposed method offers a promising solution for deepfake detection, achieving high accuracy and robustness in identifying manipulated videos.This paper proposes an innovative approach to deepfake video detection by integrating features derived from ant colony optimization–particle swarm optimization (ACO-PSO) and deep learning techniques. The method extracts ACO-PSO features from the spatial and temporal characteristics of video frames, capturing subtle patterns indicative of deepfake manipulation. These features are then used to train a deep learning classifier to distinguish between authentic and deepfake videos. Extensive experiments on benchmark datasets demonstrate the method's superior performance in terms of detection accuracy, robustness to manipulation techniques, and generalization to unseen data. The approach achieves an accuracy of 98.91% and an F1 score of 99.12%, indicating its effectiveness in deepfake detection. The integration of ACO-PSO features and deep learning enhances the precision and resilience of deepfake detection, addressing challenges in facial forgery detection and contributing to the safeguarding of digital media integrity. The method is computationally efficient and suitable for real-time applications. However, it faces limitations such as dataset size, computational complexity, robustness to new manipulation techniques, and ethical considerations. Potential solutions include expanding the dataset, optimizing the model, developing adaptability mechanisms, and establishing ethical frameworks. The study concludes that the proposed method offers a promising solution for deepfake detection, achieving high accuracy and robustness in identifying manipulated videos.