Explainable Cnn-Based Framework For Detection And Source Attribution Of Ai-Generated Images

Authors

  • 1B. Prashant,2Deevena Bhavana,3Kali Dinesh,4Kuna Sai Hari Hara Kumar,5Gutru Vamsi Author

DOI:

https://doi.org/10.62643/

Keywords:

Explainable Artificial Intelligence (XAI), Convolutional Neural Networks (CNN), AIGenerated Image Detection, Source Attribution, Deep Learning, Image Forensics, Synthetic Media Analysis, Generative Models, Digital Image Authentication, Model Interpretability.

Abstract

The rapid advancement of generative models, including Generative Adversarial Networks
(GANs) and diffusion-based models, has significantly increased the creation of AI-generated
images, raising major concerns in areas such as digital forensics, media verification, and
cybersecurity. This study presents a robust framework for detecting AI-generated images
using Convolutional Neural Networks (CNNs), which effectively learn subtle visual artifacts,
frequency-domain discrepancies, and texture patterns that differentiate synthetic images from
authentic ones. To improve interpretability, the proposed framework incorporates Explainable
Artificial Intelligence (XAI) techniques that provide visual and quantitative insights into the
model’s decision-making process, thereby enhancing transparency and user trust.
Furthermore, a Cross-Model Explainability through Source Attribution module is introduced
to compare and interpret outputs from multiple detection models, enabling identification and
attribution of the likely generative source, such as specific GAN or diffusion-based
architectures. Experimental evaluations on benchmark datasets demonstrate that the proposed
approach achieves a detection accuracy of 98.5%, showing strong generalization across
diverse AI image generation methods and reliable source attribution performance. The
integrated framework improves the reliability of AI-generated image detection while
supporting ethical and accountable use of generative AI by offering a comprehensive tool for
detection and forensic analysis

Downloads

Published

04-04-2026

How to Cite

Explainable Cnn-Based Framework For Detection And Source Attribution Of Ai-Generated Images. (2026). International Journal of Engineering Research and Science & Technology, 22(2), 811-817. https://doi.org/10.62643/