Targeted Pooled Latent-Space Steganalysis Applied to Generative Steganography, with a Fix

2025-10-14

AI Summary

Generative steganography schemes (e.g., Hu et al.) exhibit strong invisibility in pixel space but harbor statistical distinguishability in the latent space of diffusion models—a previously unaddressed security blind spot. Method: We pioneer the shift of steganalysis from pixel space to the latent space of diffusion models. Theoretically, we show that cover latent vectors follow i.i.d. Gaussian distributions, whereas stego vectors—modulated by deterministic seeds—are constrained to a hypersphere, causing significant ℓ₂-norm distribution deviation. Leveraging this insight, we propose a likelihood-ratio test under Gaussian assumptions, integrated within a pooled steganalysis framework and enhanced by randomized norm sampling. Contribution/Results: Our method achieves >99% AUC detection accuracy and fully mitigates the vulnerability: after norm randomization, stego vectors regain Gaussianity, eliminating latent-space detectability. This work uncovers a novel fragility in generative steganography and establishes a new paradigm for latent-space steganalysis.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Security and Privacy: Large-scale security measurements
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Steganographic schemes dedicated to generated images modify the seed vector in the latent space to embed a message, whereas most steganalysis methods attempt to detect the embedding in the image space. This paper proposes to perform steganalysis in the latent space by modeling the statistical distribution of the norm of the latent vector. Specifically, we analyze the practical security of a scheme proposed by Hu et. al. for latent diffusion models, which is both robust and practically undetectable when steganalysis is performed on generated images. We show that after embedding, the Stego (latent) vector is distributed on a hypersphere while the Cover vector is i.i.d. Gaussian. By going from the image space to the latent space, we show that it is possible to model the norm of the vector in the latent space under the Cover or Stego hypothesis as Gaussian distributions with different variances. A Likelihood Ratio Test is then derived to perform pooled steganalysis. The impact of the potential knowledge of the prompt and the number of diffusion steps, is also studied. Additionally, we also show how, by randomly sampling the norm of the latent vector before generation, the initial Stego scheme becomes undetectable in the latent space.

Problem

• Detects steganography in latent space not image space
• Analyzes security of generative steganographic schemes
• Models statistical distribution of latent vector norms

Innovation

• Steganalysis performed in latent space
• Modeling latent vector norm as Gaussian distributions
• Using Likelihood Ratio Test for pooled steganalysis