Deniable Steganography

TL;DR

This paper introduces the concept of deniable steganography, proposing a receiver-deniable scheme using deep neural networks that allows hiding fake messages to deceive coercive adversaries, enhancing privacy in steganographic communication.

Contribution

It pioneers the idea of deniable steganography and presents a DNN-based scheme enabling receivers to hide fake messages under coercion.

Findings

The scheme effectively embeds real and fake messages.

Experiments show high scalability and sensitivity.

The method successfully deceives coercive adversaries.

Abstract

Steganography conceals the secret message into the cover media, generating a stego media which can be transmitted on public channels without drawing suspicion. As its countermeasure, steganalysis mainly aims to detect whether the secret message is hidden in a given media. Although the steganography techniques are improving constantly, the sophisticated steganalysis can always break a known steganographic method to some extent. With a stego media discovered, the adversary could find out the sender or receiver and coerce them to disclose the secret message, which we name as coercive attack in this paper. Inspired by the idea of deniable encryption, we build up the concepts of deniable steganography for the first time and discuss the feasible constructions for it. As an example, we propose a receiver-deniable steganographic scheme to deal with the receiver-side coercive attack using deep neural networks (DNN). Specifically, besides the real secret message, a piece of fake message is also embedded into the cover. On the receiver side, the real message can be extracted with an extraction module; while once the receiver has to surrender a piece of secret message under coercive attack, he can extract the fake message to deceive the adversary with another extraction module. Experiments demonstrate the scalability and sensitivity of the DNN-based receiver-deniable steganographic scheme.