Are Deepfakes a Game Changer in Digital Images Steganography Leveraging the Cover-Source-Mismatch? - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2024

Are Deepfakes a Game Changer in Digital Images Steganography Leveraging the Cover-Source-Mismatch?

Les Deepfakes changent-ils la donne dans le domaine de la stéganographie d'images numériques en tirant parti de la disparité entre les sources ?

Arthur Méreur
  • Fonction : Collaborateur
  • PersonId : 1389294
Antoine Mallet

Résumé

This work explores the potential of synthetic media generated by AI, often referred to as Deepfakes, as a source of cover-objects for steganography. Deepfakes offer a vast and diverse pool of media, potentially improving steganographic security by leveraging cover-source mismatch, a challenge in steganalysis where training and testing data come from different sources. The present paper proposes an initial study on Deepfakes' effectiveness in the field of steganography. More precisely, we propose an initial investigation to assess the impact of Deepfakes on image steganalysis performance in an operational environment. Using a wide range of image generation models and state-of-the-art methods in steganography and steganalysis, we show that Deepfakes can significantly exploit the cover-source mismatch problem but that mitigation solutions also exist. The empirical findings can inform future research on steganographic techniques that exploit cover-source mismatch for enhanced security.
Fichier principal
Vignette du fichier
DeepStego_vFinal.pdf (579.25 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04601453 , version 1 (05-06-2024)
hal-04601453 , version 2 (17-06-2024)

Licence

Identifiants

Citer

Arthur Méreur, Antoine Mallet, Rémi Cogranne. Are Deepfakes a Game Changer in Digital Images Steganography Leveraging the Cover-Source-Mismatch?. The 19th International Conference on Availability, Reliability and Security, Jul 2024, Vienne (AUT), Austria. ⟨10.1145/3664476.3670893⟩. ⟨hal-04601453v2⟩
0 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More