Multimodal LLMs-Assisted Context-Aware Framework for Cyber Forensics

View at publisher Summarise with VIRAT
Authors: Devesh Khanna, C.R.S. Kumar
Cited by: 0
References: 28

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The emergence of semantically oriented steganography-where concealed information is interwoven within natural language, multimedia artefacts, or AI-generated material presents escalating complexities for digital forensic practitioners. Traditional detection approaches, which rely primarily on statistical or structural deviations, increasingly prove insufficient in exposing concealment strategies rooted in meaning and context. This paper presents a perspective on how multimodal large language models (LLMs) can assist forensic analysts by providing semantic and contextual reasoning capabilities. Rather than replacing established tools, LLMs can augment analytical workflows by interpreting linguistic patterns, identifying contextual anomalies, and correlating cross-modal evidence. The conceptual framework emphasizes how LLMs can act as interpretive aids to guide human judgment, offering hypothesis generation and semantic insight during complex investigations. It also reflects on key limitations, including model bias, explainability, and evidentiary reliability. This concept advocates for a human-centered integration of LLMs in digital forensicspositioning them as cognitive partners that enhance investigators’ ability to detect, interpret, and reason about concealed communication in the evolving landscape of steganography.

Fields of Study & Subjects

Digital and Cyber Forensics Advanced Steganography and Watermarking Techniques Digital Media Forensic Detection

Publication details

Journal / source
Biennial International Conference on Nascent Technologies in Engineering (ICNTE)
Pages
1–6
Publication date
16 January 2026
Language
EN