Exploring the Robustness of AI-Driven Tools in Digital Forensics: A Preliminary Study

View on arXiv ← Back to list

Authors: Silvia Lucia Sanna, Leonardo Regano, Davide Maiorca, Giorgio Giacinto

Published: 2024-12-02 10:48:53+00:00

AI Summary

This preliminary study evaluates the robustness of two AI-driven digital forensics tools, Magnet AI and Excire Photo AI, against adversarial attacks. The authors find that both tools exhibit vulnerabilities, misclassifying images and failing to reliably detect deepfakes, highlighting the need for improved robustness in AI-based forensic analysis.

Abstract

Nowadays, many tools are used to facilitate forensic tasks about data extraction and data analysis. In particular, some tools leverage Artificial Intelligence (AI) to automatically label examined data into specific categories (ie, drugs, weapons, nudity). However, this raises a serious concern about the robustness of the employed AI algorithms against adversarial attacks. Indeed, some people may need to hide specific data to AI-based digital forensics tools, thus manipulating the content so that the AI system does not recognize the offensive/prohibited content and marks it at as suspicious to the analyst. This could be seen as an anti-forensics attack scenario. For this reason, we analyzed two of the most important forensics tools employing AI for data classification: Magnet AI, used by Magnet Axiom, and Excire Photo AI, used by X-Ways Forensics. We made preliminary tests using about $200$ images, other $100$ sent in $3$ chats about pornography and teenage nudity, drugs and weapons to understand how the tools label them. Moreover, we loaded some deepfake images (images generated by AI forging real ones) of some actors to understand if they would be classified in the same category as the original images. From our preliminary study, we saw that the AI algorithm is not robust enough, as we expected since these topics are still open research problems. For example, some sexual images were not categorized as nudity, and some deepfakes were categorized as the same real person, while the human eye can see the clear nudity image or catch the difference between the deepfakes. Building on these results and other state-of-the-art works, we provide some suggestions for improving how digital forensics analysis tool leverage AI and their robustness against adversarial attacks or different scenarios than the trained one.

Key findings

Both Magnet AI and Excire Photo AI showed significant vulnerabilities. Magnet AI missed some nude images and exhibited high false positive rates in chat analysis. Excire Photo AI misidentified deepfakes as real individuals and struggled with facial recognition after cosmetic surgeries or gender-affirming procedures. The results demonstrate the need for more robust AI algorithms in digital forensics.

Approach

The researchers tested Magnet AI and Excire Photo AI using a custom dataset of images and chat logs, including deepfakes and images depicting nudity in various contexts. They assessed the tools' accuracy in classifying these images and chat content, evaluating false positives and negatives.

Datasets

A custom dataset of approximately 200 images for nudity detection (including various types of nudity and manipulated images), 200 images for face recognition (including deepfakes and images of individuals before and after cosmetic or gender-affirming surgery), and 3 chat logs containing messages and images related to drugs, weapons, and sex.

Model(s)

Proprietary AI models within Magnet AI (used by Magnet Axiom) and Excire Photo AI (used by X-Ways Forensics). Specific model architectures are not detailed.

Author countries

Italy

← Previous