Detection of Accessible Counterfeit Banknotes Using Artificial Intelligence: A Comparative Analysis of Legal and Ethical Approaches

Evolving counterfeiting techniques have turned counterfeit banknote detection into a multidimensional problem that can no longer be fully addressed through traditional security measures alone. While artificial intelligence (AI), particularly convolutional neural networks (CNNs), has emerged as a promising solution, its success remains heavily dependent on training data quality. Access to genuine images is restricted by copyright and central bank regulations, forcing a reliance on synthetic data, which often triggers domain shift and reliability issues. This study employs a comparative case analysis, treating the Turkish and international ecosystems as two primary macro-cases evaluated against legal, ethical, technical, and accessibility criteria. By establishing a causal linkage between technical errors and their resulting legal liabilities, the research explicitly differentiates between the criminal and civil consequences of misclassification. The findings indicate that focusing solely on technical accuracy is insufficient. Systemic limitations related to dataset transparency, regulatory compliance, and accessibility implementation represent common weaknesses in both settings. Consequently, the study proposes a lifecycle-anchored governance assessment rubric that aligns technical performance with legal accountability and inclusive design principles, providing a replicable evaluative instrument for the trustworthy deployment of AI in financial security applications.