We present a deep learning-based image forgery detection model that combines convolutional neural networks (CNN) with forensic techniques. Trained on data up to October 2023, the proposed model accurately detects different types of image manipulation, including subtle and sophisticated forgeries that previous models cannot detect. Extensive experiments show that our model significantly outperforms the existing state-of-the-art methods in terms of accuracy, robustness, and scalability. In addition, they can be fine-tuned and adapted to specific tasks with significantly less data compared to traditional deep learning approaches, which can have a positive impact on the carbon footprint of training deep models. It is also scalable, making it feasible to deploy the model on low-power edge devices that can enable a more sustainable digital infrastructure. These findings show that the model shows promise for real-world applications like verifying digital content and forensic analysis, enabling reliable and sustainable AI solutions.