Zero-shot Generative Linguistic Steganography
This addresses the challenge of secure and undetectable communication in linguistic steganography, though it appears incremental as it builds on existing methods with new metrics and evaluations.
The paper tackles the problem of generating stegotext that is both statistically and perceptually indistinguishable from natural text, achieving a 1.926x improvement in producing innocent and intelligible stegotext compared to other methods.
Generative linguistic steganography attempts to hide secret messages into covertext. Previous studies have generally focused on the statistical differences between the covertext and stegotext, however, ill-formed stegotext can readily be identified by humans. In this paper, we propose a novel zero-shot approach based on in-context learning for linguistic steganography to achieve better perceptual and statistical imperceptibility. We also design several new metrics and reproducible language evaluations to measure the imperceptibility of the stegotext. Our experimental results indicate that our method produces $1.926\times$ more innocent and intelligible stegotext than any other method.