AI in Digital and Mobile Forensics: A Thematic Literature Review

Abstract

The vast amount of information that modern devices produce presents a big challenge for investigators— one that Artificial Intelligence (AI) and Machine Learning (ML) seem perfectly suited to solve. Indeed, AI and ML have been used more and more in digital forensics over the past decade. To better understand this trend, we conducted a thematic literature review of 29 papers published between 2010 and 2025. These papers fell into seven categories: core ML techniques; what practitioners think; image/multimedia classification; explainable AI; robustness/adversarial issues; large language models (LLMs); and governance/legal matters. We also examined three overarching issues: how accurate these tools really are; whether they can be trusted (a question that goes to their legal admissibility); and whether they produce results that humans can understand. Based on our findings, we identified three urgent needs for the field: standardized operating procedures; public benchmarks; and systems that are not only technically sound but also legally admissible and human-friendly.