Trustworthy Visual Forensics in the Generative AI Era: Deepfakes, Biometrics, Explainable AI and Bias-Aware Analysis

The rapid advancement of generative AI is reshaping digital forensics and biometrics, enabling the creation of synthetic or manipulated digital evidence with an unprecedented level of realism and scale. This lecture examines how trustworthy AI can support the analysis of visual and biometric data, with a focus on explainability, demographic bias, and evidence reliability. Topics include deepfake detection, biometric and soft-biometric attribute analysis, and the forensic interpretation of AI-generated multimedia content. Particular emphasis is placed on transparency in decision-making and informed assessment of bias, arguing that forensic AI systems must go beyond mere detection accuracy to provide interpretable and scientifically rigorous results, suitable for digital investigations, security applications, and disinformation analysis.