The project aims to address the growing challenge of detecting AI-generated images, particularly those created using advanced diffusion models that go beyond traditional GAN-based approaches in terms of image detail and variety. While existing detection systems perform well on GAN-generated images, they struggle with the complexity of diffusion model outputs. This study attempts to create a new detection method using a One-Class SVM and a fine-tuned ResNet50 model. By leveraging and contrasting the strengths of anomaly detection and deep learning, the project targets current detection systems for detecting gaps in diffusion training sets, effectively identifying fake images generated by diffusion models without the need for large amounts of labeled data. This work not only provides a new perspective on Deepfake detection, but also contributes to the broader field of AI-generated image detection, supporting efforts to maintain social trust and information authenticity in the era of rapidly evolving generative AI technologies.