A Systematic Review of Artificial Intelligence Techniques for Evidence Analysis and Automation in Forensic Computing

Abstract

The shape of digital crime has changed, and digital forensics has not fully caught up. Evidence no longer lives in a single device. It is scattered across laptops, phones, cloud accounts, IoT systems, and the everyday platforms people use and the resulting volume is more than any human analyst can process in time. AI and ML have become the most credible response, accelerating tasks such as evidence classification, anomaly detection, malware analysis, multimedia examination, and predictive cybercrime investigation. The question is no longer whether AI helps, but where, how, and under what conditions it can be trusted inside an investigation. To answer that, this paper synthesises 24 peer-reviewed studies published between 2021 and 2026, retrieved from IEEE Xplore, ScienceDirect, SpringerLink, MDPI, and Scopus. Four themes structure the analysis: evidence classification, automated investigation, explainable AI, and ethical compliance. The evidence shows that deep learning, explainable AI, and integrated forensic frameworks can meaningfully raise the speed, accuracy, and decision-support capacity of digital investigations. At the same time, three structural problems hold the field back: the absence of standard forensic datasets, weak reproducibility, limited legal validation, and insufficient ethical governance. Therefore, AI-based digital forensics must be explainable, reliable, legally admissible, and ethically accountable.