As deepfake technology continues to evolve and proliferate, it presents significant challenges to cybersecurity, legal integrity, and personal privacy . The increasing accessibility of tools for creating synthetic media has led to a surge in the production of misleading content, including fraudulent videos that can misrepresent individuals and events. This study aims to enhance the understanding of deepfake detection by implementing a Convolutional Neural Network (CNN)- based method for a comprehensive statistical analysis of the Face- Forensics++ dataset. By examining demographic and condition-related features within this dataset, we seek to uncover patterns related to age, gender, and other demographic factors that may influence detection performance. Our findings will not only provide a detailed statistical distribution of these features but also identify potential biases in existing detection algorithms, thereby contributing to the development of more equitable and effective deepfake detection systems.