ALiSa: Acrostic Linguistic Steganography Based on BERT and Gibbs Sampling

In this letter, we propose a novel linguistic steganographic method that directly conceals a token-level secret message in a seemingly-natural steganographic text generated by the off-the-shelf BERT model equipped with Gibbs sampling. Compared with all modification based linguistic steganographic methods, the proposed method does not modify a given cover text. Instead, the proposed method utilizes the secret message to directly generate the steganographic text. Compared with mainstream generation based linguistic steganographic methods, the proposed method enables the receiver to collect the tokens of the specific positions to directly constitute the secret message, without a complex decoding process and much side information shared between the sender and the receiver. Experimental results show that the proposed method can generate fluent, highly readable steganographic texts, while enjoying pretty good anti-steganalysis ability. This work has great application potential in real-time covert communication.

Steganography [1] is a security technology that hides secret information in public carriers to realize covert communication. Compared with cryptography, it not only hides the content of secret information, but also hides the existence of secret information, demonstrating the superior security. Since its creation, text has always been an important carrier for human communication. It also makes text steganography (or linguistic steganography) have a broad application prospect. Different from digital images that may cause communication failure due to data compression, linguistic steganography enjoys the advantage of high robustness during data transmission, which better ensures the success of covert communication [2].