L2 · M = C2 Large Language Models Are Covert Channels

Conference Paper (2025)
Author(s)

Simen Gaure

S. Koffas (TU Delft - Electrical Engineering, Mathematics and Computer Science)

S. Picek

Sondre Rønjom

Research Group
Cyber Security
DOI related publication
https://doi.org/10.1109/ICASSP49660.2025.10887756 Final published version
More Info
expand_more
Publication Year
2025
Language
English
Research Group
Cyber Security
Journal title
ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Event
ICASSP (2025-04-06 - 2025-04-11), Hyderabad, India
Downloads counter
32
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Large Language Models (LLMs) are susceptible to various attacks but can also improve the security of diverse systems. However, how well do open source LLMs behave as covertext distributions to, e.g., facilitate censorship-resistant communication? In this paper, we explore open-source LLM-based covert channels. We empirically measure the security vs. capacity of two open-source LLM models (Llama-7B and GPT-2) to assess their performance as covert channels. Although our results indicate that such channels are not likely to achieve high practical bitrates, we also show that the chance for an adversary to detect covert communication is low. To ensure our results can be used with the least effort as a general reference, we employ a conceptually simple and concise scheme and only assume public models.

Files

L2_M_C2_Large_Language_Models_... (pdf)
(pdf | 0.549 Mb)
- Embargo expired in 17-04-2026
Taverne