Abstract :

The Rapid Advancement Of Generative Models Such As Generative Adversarial Networks (GANs) Has Led To The Creation Of Highly Realistic AI-generated Images That Are Often Indistinguishable From Authentic Photographs. While These Developments Have Many Beneficial Applications, They Also Pose Significant Challenges Related To Misinformation, Digital Forgery, And Privacy Violations. Therefore, Reliable Detection Of AI-generated Images Has Become An Urgent Necessity To Maintain Trust In Digital Media.In This Study, We Propose A Convolutional Neural Network (CNN)-based Approach To Accurately Distinguish AI-generated Images From Real Ones. Our Method Leverages The Powerful Feature Extraction Capabilities Of CNNs To Identify Subtle Artifacts And Inconsistencies Inherent In Synthetic Images. We Employ A Well-structured Dataset Consisting Of Real Images Sourced From Publicly Available Databases Alongside AI-generated Images Created By State-of-the-art GAN Models.To Enhance The Interpretability Of Our Models Decisions, We Integrate Explainable AI (XAI) Techniques, Specifically Grad-CAM And SHAP, Which Provide Visual And Quantitative Explanations For The CNNs Predictions. These Techniques Reveal The Critical Regions And Features In The Images That The Model Uses To Differentiate Between Real And AI-generated Content. This Interpretability Not Only Aids In Understanding Model Behavior But Also Increases User Trust And Model Transparency.Extensive Experiments Demonstrate That Our CNN Model Achieves High Accuracy And Robustness In Classifying AI-generated Images Across Various Datasets. The XAI Visualizations Further Confirm That The Model Focuses On Meaningful Image Regions, Such As Unnatural Textures Or Artifacts Often Introduced By Generative Algorithms. These Findings Highlight The Effectiveness Of Combining Deep Learning With Explainability For Digital Image Forensics.Despite Promising Results, The Approach Has Limitations, Including Potential Vulnerability To Adversarial Examples And Generalization Challenges With Novel GAN Architectures. Future Work Will Explore Ensemble Models, Multi-modal Data Integration, And Real-time Detection Frameworks To Address These Challenges And Improve Detection Reliability.

Published:

30-7-2026

Issue:

Vol. 26 No. 7 (2026)

Page Nos:

1248-1254

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite