DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural Architectures

Modern linguistic steganography faces the fundamental challenge of balancing embedding capacity with detection resistance, particularly against advanced AI-based steganalysis. This paper presents DeepStego, a novel steganographic system leveraging GPT-4-omni’s language modeling capabilities for secu...

Full description

Saved in:
Bibliographic Details
Main Authors: Oleksandr Kuznetsov, Kyrylo Chernov, Aigul Shaikhanova, Kainizhamal Iklassova, Dinara Kozhakhmetova
Format: Article
Language:English
Published: MDPI AG 2025-04-01
Series:Computers
Subjects:
Online Access:https://www.mdpi.com/2073-431X/14/5/165
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850127361112014848
author Oleksandr Kuznetsov
Kyrylo Chernov
Aigul Shaikhanova
Kainizhamal Iklassova
Dinara Kozhakhmetova
author_facet Oleksandr Kuznetsov
Kyrylo Chernov
Aigul Shaikhanova
Kainizhamal Iklassova
Dinara Kozhakhmetova
author_sort Oleksandr Kuznetsov
collection DOAJ
description Modern linguistic steganography faces the fundamental challenge of balancing embedding capacity with detection resistance, particularly against advanced AI-based steganalysis. This paper presents DeepStego, a novel steganographic system leveraging GPT-4-omni’s language modeling capabilities for secure information hiding in text. Our approach combines dynamic synonym generation with semantic-aware embedding to achieve superior detection resistance while maintaining text naturalness. Through comprehensive experimentation, DeepStego demonstrates significantly lower detection rates compared to existing methods across multiple state-of-the-art steganalysis techniques. DeepStego supports higher embedding capacities while maintaining strong detection resistance and semantic coherence. The system shows superior scalability compared to existing methods. Our evaluation demonstrates perfect message recovery accuracy and significant improvements in text quality preservation compared to competing approaches. These results establish DeepStego as a significant advancement in practical steganographic applications, particularly suitable for scenarios requiring secure covert communication with high embedding capacity.
format Article
id doaj-art-467572d5c4a54c239fd856f8d77957ce
institution OA Journals
issn 2073-431X
language English
publishDate 2025-04-01
publisher MDPI AG
record_format Article
series Computers
spelling doaj-art-467572d5c4a54c239fd856f8d77957ce2025-08-20T02:33:42ZengMDPI AGComputers2073-431X2025-04-0114516510.3390/computers14050165DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural ArchitecturesOleksandr Kuznetsov0Kyrylo Chernov1Aigul Shaikhanova2Kainizhamal Iklassova3Dinara Kozhakhmetova4Department of Theoretical and Applied Sciences, eCampus University, Via Isimbardi 10, 22060 Novedrate, ItalyDepartment of Intelligent Software Systems and Technologies, School of Computer Science and Artificial Intelligence, V. N. Karazin Kharkiv National University, 4 Svobody Sq., 61022 Kharkiv, UkraineDepartment of Information Security, L.N. Gumilyov Eurasian National University, Satpayev 2, Astana 010008, KazakhstanDepartment of Information and Communication Technologies, Manash Kozybayev North Kazakhstan University, Pushkin Str., 86, Petropavlovsk 150000, KazakhstanHigher School of Artificial Intelligence and Construction, Shakarim University, St. Glinka, 20A, Semey 071412, KazakhstanModern linguistic steganography faces the fundamental challenge of balancing embedding capacity with detection resistance, particularly against advanced AI-based steganalysis. This paper presents DeepStego, a novel steganographic system leveraging GPT-4-omni’s language modeling capabilities for secure information hiding in text. Our approach combines dynamic synonym generation with semantic-aware embedding to achieve superior detection resistance while maintaining text naturalness. Through comprehensive experimentation, DeepStego demonstrates significantly lower detection rates compared to existing methods across multiple state-of-the-art steganalysis techniques. DeepStego supports higher embedding capacities while maintaining strong detection resistance and semantic coherence. The system shows superior scalability compared to existing methods. Our evaluation demonstrates perfect message recovery accuracy and significant improvements in text quality preservation compared to competing approaches. These results establish DeepStego as a significant advancement in practical steganographic applications, particularly suitable for scenarios requiring secure covert communication with high embedding capacity.https://www.mdpi.com/2073-431X/14/5/165linguistic steganographyGPT modelsnatural language processinginformation hidingtext generationsemantic embedding
spellingShingle Oleksandr Kuznetsov
Kyrylo Chernov
Aigul Shaikhanova
Kainizhamal Iklassova
Dinara Kozhakhmetova
DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural Architectures
Computers
linguistic steganography
GPT models
natural language processing
information hiding
text generation
semantic embedding
title DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural Architectures
title_full DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural Architectures
title_fullStr DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural Architectures
title_full_unstemmed DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural Architectures
title_short DeepStego: Privacy-Preserving Natural Language Steganography Using Large Language Models and Advanced Neural Architectures
title_sort deepstego privacy preserving natural language steganography using large language models and advanced neural architectures
topic linguistic steganography
GPT models
natural language processing
information hiding
text generation
semantic embedding
url https://www.mdpi.com/2073-431X/14/5/165
work_keys_str_mv AT oleksandrkuznetsov deepstegoprivacypreservingnaturallanguagesteganographyusinglargelanguagemodelsandadvancedneuralarchitectures
AT kyrylochernov deepstegoprivacypreservingnaturallanguagesteganographyusinglargelanguagemodelsandadvancedneuralarchitectures
AT aigulshaikhanova deepstegoprivacypreservingnaturallanguagesteganographyusinglargelanguagemodelsandadvancedneuralarchitectures
AT kainizhamaliklassova deepstegoprivacypreservingnaturallanguagesteganographyusinglargelanguagemodelsandadvancedneuralarchitectures
AT dinarakozhakhmetova deepstegoprivacypreservingnaturallanguagesteganographyusinglargelanguagemodelsandadvancedneuralarchitectures