Provably Secure Covert Messaging Using Image-based Diffusion Processes

2025-03-13

AI Summary

This work addresses the security and robustness challenges of steganography in image diffusion models. We propose a novel steganographic method that embeds messages directly into the initial latent space—without altering the prior distribution of latent variables—and provide, for the first time, a computational indistinguishability proof against arbitrary polynomial-time adversaries. Our approach integrates provably secure probabilistic design, a robust inverse transformation mechanism, and joint error-correction optimization. It achieves high payload capacity (>100 bits), strong undetectability (AUC < 0.52 against deep neural network detectors), and resilience to common distortions (e.g., 30% cropping, JPEG compression at quality factor 30). Experimental results demonstrate that our method attains an optimal trade-off among capacity, robustness, and reliability—significantly outperforming existing steganographic schemes based on pixel-level manipulation or post-hoc latent-space processing.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

We consider the problem of securely and robustly embedding covert messages into an image-based diffusion model's output. The sender and receiver want to exchange the maximum amount of information possible per diffusion sampled image while remaining undetected. The adversary wants to detect that such communication is taking place by identifying those diffusion samples that contain covert messages. To maximize robustness to transformations of the diffusion sample, a strategy is for the sender and the receiver to embed the message in the initial latents. We first show that prior work that attempted this is easily broken because their embedding technique alters the latents' distribution. We then propose a straightforward method to embed covert messages in the initial latent {em without} altering the distribution. We prove that our construction achieves indistinguishability to any probabilistic polynomial time adversary. Finally, we discuss and analyze empirically the tradeoffs between embedding capacity, message recovery rates, and robustness. We find that optimizing the inversion method for error correction is crucial for reliability.

Problem

• Secure embedding of covert messages in image diffusion models.
• Maximizing information exchange while avoiding detection by adversaries.
• Ensuring robustness and reliability in message recovery and embedding.

Innovation

• Embeds covert messages in initial latents
• Maintains latent distribution integrity
• Optimizes inversion for error correction