Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{198107,
        author = {Arul Kumaran P and Subramanian E.K and Harshitha R and Harish Raghavendra R},
        title = {End-to-End AI Watermarking Framework for Generative Models Using MVEPC Method},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {11672-11680},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=198107},
        abstract = {This study introduces an invisible watermarking framework based on deep learning that is intended to ensure ownership verification and tamper-resistant authentication for AI-generated images. The proposed system integrates Multi-Variant Error Patch Coding (MVEPC) to embed watermark fragments into latent feature representations, enabling high imperceptibility and strong robustness against common image distortions. The encoder distributes redundant micro-patches across the image, while the decoder reconstructs the watermark even when 40–60% of the image is manipulated. Experimental evaluation demonstrates that the system achieves PSNR values above 38 dB, SSIM above 0.98, and high watermark recovery accuracy under compression, cropping, noise addition, and social media recompression. This approach provides a secure, resilient, and scalable solution for digital rights protection in modern generative AI environments.},
        keywords = {Invisible watermarking, MVEPC, deep learning, tamper detection, generative AI, latent embedding, image authentication, robustness.},
        month = {April},
        }