Cognitive Cyber Crimes in the Era of Artificial Intelligence

6Developing AI-Enhanced Forensic Tools to Combat Cognitive Cybercrimes

Priya Matta¹*, Atika Gupta², R.K. Saini³, Sopan Talekar⁴ and Manjusha Tatiya⁵

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

²Graphic Era Hill University, Dehradun, India

³School of Computing, DIT University, Dehradun, India

⁴MVPS Karmaveer Adv. Baburao Ganpatrao Thakare College of Engineering, Nashik, India

⁵Department of AI DS, Indira College of Engineering and Management, Pune, India

Abstract

Cognitive cybercrimes—crimes aimed at manipulating human cognition through social engineering, deepfakes, misinformation, and synthetic identities—pose significant risks in today’s digital ecosystem. This research introduces artificial intelligence–enhanced forensic tools that integrate machine learning and natural language processing (NLP) to detect, analyze, and mitigate cognitive cyberthreats. Utilizing the RealNews dataset (45 million tokens, 120K labeled articles), the FaceForensics++ dataset (1,000+ manipulated videos), and the SYN-ID dataset (synthetically generated identity profiles), we apply an ensemble approach combining BERT-based NLP classification, XceptionNet for deepfake detection, and gradient boosted decision trees for identity fraud classification. Our hybrid model achieved an accuracy of 97.6% in identifying misinformation, 96.3% for deepfake detection, and 95.1% in synthetic identity fraud detection. New performance metrics such as the Cognitive Threat Precision ...