Cognitive Cyber Crimes in the Era of Artificial Intelligence

7Social Engineering in the AI Era: Exploiting Human Perception for Threat

Sunil Parihar¹*, Ankit Jain², Rahul Verma², Rupa Rajoriya² and Kapil Bhardwaj¹

¹SAIT Indore, CSE, Indore, India

²PITS Ujjain, CSE, Indore, India

Abstract

In the evolving digital landscape, social engineering attacks have become increasingly sophisticated, particularly with the integration of artificial intelligence (AI) that manipulates human perception for malicious intent. This research investigates AI-enabled social engineering techniques and proposes a novel hybrid detection model—Perception-Aware Threat Analysis Framework (PTAF)—that integrates deep deceptive pattern recognition (DDPR) and human cognitive vulnerability estimation. We utilized the Social Engineering Detection Dataset, containing over 50,000 labeled records of phishing, pretexting, and baiting scenarios enriched with behavioral and cognitive cues. Experimental evaluation was performed using advanced classifiers including BERT, BiLSTM, and a custom transformer-DDPR model. The proposed PTAF model achieved a detection accuracy of 98.91%, precision of 97.83%, F1 score of 98.22%, and false-positive rate of 1.09%, outperforming traditional natural language processing–based baselines. This study emphasizes AI’s dual-edged nature in both perpetuating and preventing social engineering, calling for enhanced perceptual-cognitive defense systems.

Keywords: AI-enabled deception, social engineering detection, human perception exploitation, cognitive ...