Cognitive Cyber Crimes in the Era of Artificial Intelligence

26Fraud and Misreporting Risks in Executive Pay: A Cognitive AI Detection Framework by Algorithmic Governance

Neha Aggarwal*, Kriti Bhaswar Singh, Devbrath and Kuntal Dutta

Department of Commerce and Financial Studies, Central University of Jharkhand, Jharkhand, India

Abstract

Executive compensation often involves complex contractual structures that create opportunities for fraud, misreporting, and governance failures. This study proposes a robust detection model that leverages cognitive artificial intelligence for identifying anomalous pay practices. Using the ExecuComp dataset (1992–2022), which contains detailed disclosures on executive compensation, firm performance, and governance attributes, the framework integrates Hybrid Cognitive Attention Networks—a novel fusion of transformer-based contextual embeddings with gradient-boosted decision trees for anomaly scoring. The methodology includes data preprocessing through normalization, feature extraction of pay-performance sensitivity, board independence metrics, and compensation structure ratios, followed by anomaly detection calibrated with adaptive governance thresholds. Experimental results demonstrate that the framework achieves 94.8% precision, 92.3% recall, and an F1 score of 93.5%, significantly outperforming conventional machine learning baselines such as random forests and support vector machine. Moreover, the cognitive interpretability module highlights the disproportionate use of stock options and deferred bonuses ...