Auto-Stega: An Agent-Driven System for Lifelong Strategy Evolution in LLM-Based Text Steganography

TL;DR

Auto-Stega presents an agent-driven, self-evolving framework for text steganography that continuously discovers, composes, and updates embedding strategies at inference time within a closed loop. It introduces PC-DNTE as a plug-and-play encoder to preserve the base distribution at high embedding rates, enabling stronger security and imperceptibility without retraining. Across News, Movie, and Tweet corpora, Auto-Stega outperforms state-of-the-art baselines, achieving a 42.2% improvement in perplexity and a 1.6% gain in anti-steganalysis, with human evaluators noting superior fluency and coherence. The approach consolidates lifelong strategy evolution, a strategy library, and plug-and-play high-rate mapping to deliver efficient, covert communication and offers avenues toward multilingual and multimodal extensions.

Abstract

With the rapid progress of LLMs, high quality generative text has become widely available as a cover for text steganography. However, prevailing methods rely on hand-crafted or pre-specified strategies and struggle to balance efficiency, imperceptibility, and security, particularly at high embedding rates. Accordingly, we propose Auto-Stega, an agent-driven self-evolving framework that is the first to realize self-evolving steganographic strategies by automatically discovering, composing, and adapting strategies at inference time; the framework operates as a closed loop of generating, evaluating, summarizing, and updating that continually curates a structured strategy library and adapts across corpora, styles, and task constraints. A decoding LLM recovers the information under the shared strategy. To handle high embedding rates, we introduce PC-DNTE, a plug-and-play algorithm that maintains alignment with the base model's conditional distribution at high embedding rates, preserving imperceptibility while enhancing security. Experimental results demonstrate that at higher embedding rates Auto-Stega achieves superior performance with gains of 42.2\% in perplexity and 1.6\% in anti-steganalysis performance over SOTA methods.

Figures (9)

• Figure 1: The pipeline of Auto-Stega.
• Figure 2: Lifelong evolution of steganographic strategies.
• Figure 3: The overall framework of Auto-Stega.
• Figure 4: Steganographic strategy library construction: from warm-up to lifelong learning.
• Figure 5: The results of the anti-steganalysis performance across low and high embedding rates.