SparSamp: Efficient Provably Secure Steganography Based on Sparse Sampling

2025-03-25

AI Summary

This work addresses the longstanding trade-off between security and efficiency in generative steganography. We propose a provably secure steganographic framework based on sparse sampling. Methodologically, we introduce a novel sparse random sampling mechanism that preserves the original output distribution of generative models (e.g., GPT-2, DDPM, WaveRNN) while replacing only the sampling layer with O(1) computational overhead; message-guided pseudorandom number generation enables key-controllable embedding. Theoretically, our framework guarantees statistical indistinguishability under standard cryptographic assumptions—ensuring rigorous, provable security. Experimentally, it achieves up to 9223 bps embedding rate across text, image, and audio modalities without degrading generation quality or inference latency. Moreover, it supports plug-and-play deployment across diverse generative architectures.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation
• Computer Vision: Generative Adversarial Networks (GANs) for Vision

Application Category

• Security and Privacy: Large-scale security measurements
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Steganography embeds confidential data within seemingly innocuous communications. Provable security in steganography, a long-sought goal, has become feasible with deep generative models. However, existing methods face a critical trade-off between security and efficiency. This paper introduces SparSamp, an efficient provably secure steganography method based on sparse sampling. SparSamp embeds messages by combining them with pseudo-random numbers to obtain message-derived random numbers for sampling. It enhances extraction accuracy and embedding capacity by increasing the sampling intervals and making the sampling process sparse. SparSamp preserves the original probability distribution of the generative model, thus ensuring security. It introduces only $O(1)$ additional complexity per sampling step, enabling the fastest embedding speed without compromising generation speed. SparSamp is designed to be plug-and-play; message embedding can be achieved by simply replacing the sampling component of an existing generative model with SparSamp. We implemented SparSamp in text, image, and audio generation models. It can achieve embedding speeds of up to 755 bits/second with GPT-2, 5046 bits/second with DDPM, and 9,223 bits/second with WaveRNN.

Problem

• Balancing security and efficiency in steganography methods
• Enhancing embedding capacity and extraction accuracy in steganography
• Maintaining generative model speed while embedding messages securely

Innovation

• Sparse sampling for secure steganography
• Plug-and-play with existing generative models
• O(1) complexity maintains fast embedding speed