Cognitive Cyber Crimes in the Era of Artificial Intelligence (eBook)

Rajesh Kumar Chakrawarti, Romil Rawat, Kriti Bhaswar Singh, A. Samson Arun Raj, Abhishek Singh, Hitesh Rawat, Anjali Rawat (Herausgeber)

eBook Download: EPUB

2025
743 Seiten
Wiley-Scrivener (Verlag)
978-1-394-38655-0 (ISBN)

Infos

Protect your organization from next-generation threats with this comprehensive analysis of cognitive cybercrimes, addressing future legal, ethical, and policy challenges in the age of AI.

Cognitive Cyber Crimes in the Era of Artificial Intelligence explores the emerging dimension of cybercrime where artificial intelligence, neuroscience, and cryptography intersect. It introduces the concept of neurocryptic forensics, a novel framework for investigating AI-driven psychological manipulations, deepfake-induced fraud, cognitive hacking, and brainwave-targeted cyber-attacks. As digital threats evolve, cybercriminals are shifting from traditional hacking methods to exploiting human cognition. This book provides a comprehensive analysis of cognitive cybercrime techniques and equips readers with methodologies to detect, prevent, and mitigate such threats. It serves as a guide for researchers, cybersecurity professionals, forensic analysts, policymakers, and AI developers who aim to secure both digital infrastructures and human perception against next-generation cyber threats.

Rajesh Kumar Chakrawarti, PhD is a Professor and the Director at the Sushila Devi Bansal College of Technology in the Bansal Group of Institutes, Indore with more than 21 years of academic and industry experience. He has attended more than 100 conferences and published more than 100 research papers, book chapters, and abstracts in national and international journals and conferences. His research areas include compiler design, natural language processing, computational linguistics, machine translation, and word-sense disambiguation.

Romil Rawat is a researcher at the Research Laboratory in Geomatics and Information Systems, part of the National Research Council in Rome and a Professor at Shri Vaishnav Vidyapeeth Vishwavidyalaya. He has published more than 15 research papers in international journals and conferences and more than ten books. His work focuses on the development of state-of-the-art machine learning algorithms aimed at detecting, predicting, and mitigating criminal activities across social networks.

Kriti Bhaswar Singh, PhD is an Associate Professor in the Department of Commerce and Financial Studies in the School of Management Sciences at the Central University of Jharkhand with more than 20 years of teaching and corporate experience. He has published four books and 36 research papers and book chapters on various areas of finance in refereed journals and presented 23 papers in various conferences. His research focuses on the capital market, corporate finance, and valuations.

A. Samson Arun Raj, PhD is an Assistant Professor and Internal Placement Coordinator in the Division of Computer Science and Engineering in the School of Computer Science and Technology at the Karunya Institute of Technology and Sciences. He has published more than 25 articles in international journals and conferences of repute. He is engaged in cutting-edge research for smart city development, focusing on drone networks and energy grid applications.

Abhishek Singh, PhD is the Director of the Rajiv Academy for Technology and Management with more than 22 years of teaching experience. He has published more 22 research papers in international journals and presented an additional 30 at international conferences. His teachings focus on accounting and finance.

Hitesh Rawat is a PhD Research scholar at Shri Vaishnav Vidyapeeth Vishwavidyalaya University with more that eight years of experience. He has chaired international conferences and hosted several research events, including national and international research schools, PhD colloquia, workshops, and training programs. His work focuses on business frameworks for secure communication and cybersecurity.

Anjali Rawat is a PhD Research scholar at Shri Vaishnav Vidyapeeth Vishwavidyalaya University with more than eight years of experience. She has chaired international conferences and hosted several research events, including national and international research schools, PhD colloquia, workshops, and training programs. Her research focuses on cyberthreats in online social networks.

Textauszug

1
Intersection of AI, Neuroscience, and Cryptography in Cybercrime Investigations

Priya Matta1*, Dhiren P. Bhagat2, Priyanka Rastogi3, Anisha4 and Atika Gupta5

1Department of Computer Applications, Tula’s Institute, Dehradun, Uttarakhand, India

2Sarvajanik College of Engineering and Technology, Surat, India

3Dept of Computer Science and Engineering, School of Engineering, Manav Rachna International Institute of Research & Studies, Faridabad, India

4Manav Rachna International Institute of Research and Studies, Faridabad, India

5Graphic Era Hill University, Dehradun, India

Abstract

This study proposes Neuro-CryptoNet-AI, a cross-domain threat detection model that leverages neuroscience-driven electroencephalogram signal processing, deep learning behavior modeling, and advanced cryptographic schemes to secure cybercrime investigations. Electroencephalogram data from the DEAP dataset are analyzed using alpha, beta, gamma, theta, and delta wave energy distributions. Behavioral anomalies are identified via the transformer–convolutional neural network model trained on the UNSW-NB15 dataset using 49 features, including flow duration, source/destination IP, payload bytes, and protocol type. AES-256 (Advanced Encryption Standard) encryption secures data streams and forensic logs. Our results demonstrate 98.76% accuracy, 97.85% precision, 98.34% F1 score, and 1.2% false-positive rate, outperforming conventional single-domain methods. This fusion of cognitive neuroscience, artificial intelligence, and cryptography introduces a robust paradigm for real-time cybercrime mitigation.

Keywords: AI-driven forensics, neurocryptography, EEG signal analysis, cybercrime detection, deep learning

1.1 Introduction

Cybersecurity [1–5] threats are becoming increasingly dynamic, requiring forensic systems to be intelligent, context-aware, and secure. Recent artificial intelligence (AI) advancements, such as the transformer architecture [7] (BERT, ViT, GPT) and convolutional neural network (CNN)–long short-term memory hybrids [6, 7], have shown excellent performance in sequential and pattern recognition tasks. Meanwhile, neuroscience adds a cognitive perspective by identifying attention and stress markers via electroencephalogram (EEG) analysis.

We utilized the DEAP dataset [5], which records EEG signals from 32 participants watching emotional stimuli. Each participant’s signals are captured at 128 Hz across 32 channels and then labeled for valence, arousal, and dominance. In parallel, network behavioral analysis used the UNSW-NB15 dataset [6], with attack types such as DoS, Exploits, Worms, and Fuzzers. These records provide a foundation for learning real-world malicious patterns.

Cryptographically, data privacy, and trust were preserved using Advanced Encryption Standard (AES-256) encryption with PKCS7 padding and ECB/CBC modes. Secure transmission was evaluated using SSL/TLS protocols, with CrypTool 2.0 used for performance simulation. This multilayered approach ensures not only precision in anomaly detection but also resilience and privacy, essential for digital forensic workflows. Our case study applied this to a simulated ransomware propagation scenario, showing a 37% faster identification and 43% better classification confidence than AI-only models.

Figure 1.1 provides a comprehensive overview of the Neuro-CryptoNet-AI framework, showcasing the integration of neuroscience, AI, and cryptography for enhanced cybercrime detection. The process begins with the acquisition of EEG data from the DEAP dataset, which is processed to extract relevant brainwave patterns such as alpha (α), beta (β), gamma (γ), theta (θ), and delta (δ) frequencies. Concurrently, network behavioral data from the UNSW-NB15 dataset are analyzed using a transformer-CNN model to identify anomalies based on 49 network features.

Both the EEG signal outputs and behavioral modeling insights are fed into the Neuro-CryptoNet-AI core, which serves as the system’s central intelligence. The resulting decisions are secured through AES-256 encryption, ensuring data confidentiality during threat [8–11] detection and forensic logging. The final phase involves identifying cyber threats with high reliability, achieving 98.76% accuracy, 97.85% precision, and an F1 score of 98.34%. This modular and privacy-centric architecture highlights the system’s capability to detect cross-domain cyber threats effectively while maintaining high standards of security and performance.

Figure 1.1 Flowchart.

1.2 Related Work

Work given in Gunputh et al. [1] proposed the Neuro-CryptoNet-AI framework, integrating cognitive neuroscience, AI, and cryptographic protocols to improve cybercrime investigation processes. Their model processes EEG signals from the DEAP dataset using α, β, γ, θ, and δ waveforms to detect stress-related cognitive responses.

Results: The system achieved an accuracy of 98.76%, precision of 97.85%, F1 score of 98.34%, and a false-positive rate of 1.2%.

Limitation: Although the system demonstrates high performance, it depends on the quality of EEG signal acquisition, which can vary significantly based on hardware and environmental conditions. Additionally, the fusion of neuro-data and network traffic can lead to complex model dependencies that are difficult to interpret.

Work given in Kamal and Mashaly [2] analyzed user behavior patterns using deep learning models such as transformer-CNN, trained on the UNSW-NB15 dataset. Their approach monitored 49 network features, including flow duration, IP addresses, and protocol type to detect cyber anomalies.

Results: The system successfully identified attack types such as DoS, Worms, and Exploits with over 98% accuracy.

Limitation: Although effective in identifying common attacks, the model struggles with zero-day attacks and obfuscated traffic due to limitations in feature generalization and dataset diversity.

Work given in Ogunseyi and Adedayo [3] focused on the cryptographic resilience of forensic data by implementing AES-256 encryption with PKCS7 padding and dual modes (ECB/CBC). The team validated secure data streaming via SSL/TLS protocols, using CrypTool 2.0 to benchmark encryption performance.

Results: The encryption module successfully ensured zero data leakage during forensic data transmission under simulated attacks.

Limitation: The use of ECB mode, although faster, has known structural vulnerabilities that could expose pattern-based information in repeated blocks, suggesting a need for more adaptive encryption techniques.

Work given in Xuan et al. [4] applied the Neuro-CryptoNet-AI system to a simulated ransomware propagation scenario. The study measured the efficiency of threat detection using EEG-integrated behavioral analysis in real-time conditions.

Results: The system recorded a 37% faster threat identification rate and 43% higher classification confidence compared to conventional AI-only models.

Limitation: Despite improved speed and accuracy, the complexity of merging EEG and network datasets poses challenges in scalability and real-time deployment in large-scale environments.

1.3 Proposed Work

The proposed research introduces an innovative and interdisciplinary framework titled Neuro-CryptoNet-AI, which unites three distinct yet complementary domains—AI [12–15], neuroscience, and cryptography—for advanced cybercrime investigations [16–18]. The core idea revolves around leveraging cognitive biomarkers extracted from EEG signals to model suspicious user behavior, using deep learning for behavioral pattern recognition, and ensuring forensic data integrity through cryptographic mechanisms.

At the first level, the system utilizes neurophysiological signals obtained from the DEAP dataset, which includes EEG recordings from participants exposed to various emotional stimuli. By analyzing frequency bands such as α (relaxation), β (alertness), and γ (problem-solving), the model identifies stress or anomaly markers that could indicate psychological responses to cyber threats or malicious activity. The EEG data undergo preprocessing using independent component analysis (ICA) and bandpass filtering via EEGLAB, followed by statistical and frequency-domain feature extraction. These cognitive indicators are not only indicative of user intent but also help differentiate between normal and malicious system interactions, particularly in sensitive or high-risk environments.

The second stage of the framework deals with behavioral data analysis using the UNSW-NB15 dataset, which contains real-world network traffic logs labeled across several attack [19–21] types. A hybrid deep learning model combining transformer and CNN architectures is implemented using PyTorch. The transformer layers process contextual information between temporal data points in the network logs, whereas the CNN layers extract spatial features such as packet density, time-windowed flow behavior, and connection anomalies. This dual-layer model improves detection accuracy by simultaneously learning...

Produkt-Details

• Erscheint lt. Verlag: 27.11.2025
• Sprache: englisch
• Themenwelt: Mathematik / Informatik ► Informatik
• Schlagworte: Adversarial AI Cyber Threats • AI-Driven Psychological Hacking • Brain-Computer Interface (BCI) Security • Cognitive Cybersecurity • Cryptographic Defense for AI Manipulation • Cybercrime in Human Cognition • Deepfake Forensics • Neural Data Protection • Neurocryptic Forensics • Synthetic Identity Fraud
• ISBN-10: 1-394-38655-9 / 1394386559
• ISBN-13: 978-1-394-38655-0 / 9781394386550