Beyond the Pixel Veil: Forensic Analysis of IDAT Signatures and Generator-Specific Artifacts in AI-Generated Images
DOI:
https://doi.org/10.15680/IJCTECE.2026.0902020Keywords:
Deepfake detection, IDAT forensics, generator fingerprinting, diffusion models, metadata analysis, real-time detection.Abstract
The swift democratization of generative AI, especially diffusion-based systems like DALL-E, Stable Diffusion, and Midjourney, has far outstripped the progress made in developing effective detection techniques (Chaniporn Thampanichwat et al., 2025). In 2026, deepfake images cannot only fool the eye but are also increasingly evasive to standard detection mechanisms (Singh & Dhumane, 2025). This paper offers a technical evaluation of three potential paradigms for detecting deepfake images: human perception, commercial AI-detectors, and low-level metadata and forensic analysis (Almutairi & Elgibreen, 2022). Based on our controlled data set (n=4; two authentic images from legitimate sources, two artificial images from two different generators: ChatGPT and Perchance), we evaluate the effectiveness of five commercial detection engines and three forensic analyzers (Wireshark, Foto-Forensics, Autopsy). Our key technical findings include: (1) AI-generated images have an order of magnitude more IDAT chunks and IDAT signatures, which are nine times as large (byte-wise) as authentic images; (2) generator-specific fingerprints can be identified (Foto-Forensics could distinguish ChatGPT images); and (3) commercial detectors give nearly random performance estimates (range from 0% to 100%). We present a lightweight heuristic detection system based on IDAT thresholds (threshold = 15 chunks, file size => 2MB). We conclude with a discussion about applications for real-time detection systems as well as the necessity for 2026 benchmarking criteria
References
1. Almutairi, Z., & Elgibreen, H. (2022). A Review of Modern Audio Deepfake Detection Methods: Challenges and Future Directions. Algorithms, 15(5), 155. https://doi.org/10.3390/a15050155
2. Chaniporn Thampanichwat, Tarid Wongvorachan, Limpasilp Sirisakdi, Pornteera Chunhajinda, Suphat Bunyarittikit, & Rungroj Wongmahasiri. (2025). Mindful Architecture from Text-to-Image AI Perspectives: A Case Study of DALL-E, Midjourney, and Stable Diffusion. Buildings, 15(6), 972–972. https://doi.org/10.3390/buildings15060972
3. Didem Pehlivanoglu, Zhu, M., Zhen, J., Gagnon-Roberge, A. A., Kern, R. K., Woodard, D., Cahill, B. S., & Ebner, N. C. (2026). Is this real? Susceptibility to deepfakes in machines and humans. Cognitive Research Principles and Implications, 11(1), 3–3. https://doi.org/10.1186/s41235-025-00700-y
4. Ginzburg-Ganz, E., Horodi, E. D., Shadafny, O., Uri Savir, Ram M., & Levron, Y. (2025). Statistical Foundations of Generative AI for Optimal Control Problems in Power Systems: Comprehensive Review and Future Directions. Energies, 18(10), 2461–2461. https://doi.org/10.3390/en18102461
5. Goyal, M., & Mahmoud, Q. H. (2024). A Systematic Review of Synthetic Data Generation Techniques Using Generative AI. Electronics, 13(17), 3509. https://doi.org/10.3390/electronics13173509
6. Kamble, V. B., & Dr. Nilesh J. Uke. (2024, November 30). Image tampering detection: a review of multi-technique approach from traditional to deep learning. https://doi.org/10.71058/jodac.v8i11024
7. Modak, S., Saltık, A. O., & Stein, A. (2026). Evaluating model quantization in a GenAI-enhanced weed detection pipeline. Journal of Systems Architecture, 175, 103755. https://doi.org/10.1016/j.sysarc.2026.103755
8. Salih, S., Husain, O., Mohamed Almohamedh, R., Tajelsier, H., Hassan Abdalla Hashim, A., Elshafie, H., & Motwakel, A. (2025). From Ideation to Execution: Unleashing the Power of Generative AI in Modern Digital Marketing and Customer Engagement- A Systematic Literature Review and Case Study. Array, 100630. https://doi.org/10.1016/j.array.2025.100630
9. Shah, J., Gromis, M., & Pinto, R. (2024, December 18). Enhancing Diffusion Models for High-Quality Image Generation. https://doi.org/10.48550/arXiv.2412.14422
10. Singh, S., & Dhumane, A. (2025). Unmasking digital deceptions: An integrative review of deepfake detection, multimedia forensics, and cybersecurity challenges. MethodsX, 15, 103632. https://doi.org/10.1016/j.mex.2025.103632
11. Thunuguntla, A., Tadepalli, P., & Raffa, G. (2025). Defenses Against Adversarial Attacks on Object Detection: Methods and Future Directions. Information, 16(11), 1003. https://doi.org/10.3390/info16111003
12. Yang, M., Goenawan, G. J., Wang, H., Qin, H., Xu, C., Yang, Y., Fang, F., Sun, Y., Lim, J.-H., & Zhu, H. (2026). Your AI-Generated Image Detector Can Secretly Achieve SOTA Accuracy, If Calibrated. ArXiv.org. https://arxiv.org/abs/2602.01973
13. Yang, P., Zhou, C., Baracchi, D., Shullani, D., Zou, Y., & Piva, A. (2026). Forensic Analysis for Source Camera Identification from EXIF Metadata. Journal of Imaging, 12(3), 110. https://doi.org/10.3390/jimaging12030110
14. Yerzhanuly, M. (2025). Deepfake Geography: Detecting AI-Generated Satellite Images. Arxiv.org. https://arxiv.org/html/2511.17766v1
15. Zhang, Y. (2025). Comparing AI and humans’ ability to recognize AI-generated images. Scholarly Review Journal, Spring 2025(12). https://doi.org/10.70121/001c.132244
