Deepfake Detection System for Facial Evidence Verification in Criminal Justice and Its Legal and Ethical Implications

Deepfake technology poses a significant threat to the integrity of facial evidence in criminal justice. The existence of highly realistic manipulated videos and images raises concerns about the authenticity and admissibility of such evidence in court. This research proposes a deepfake detection system for facial evidence verification to combat the threat of deepfake technology on facial evidence in criminal justice. Our experimental results indicate that the introduction of confidence thresholds, prediction timestamps, and heatmaps of individual frames of the suspected video can be useful and practical in the criminal justice domain preventing wrongful convictions of innocent victims of deepfake technology manipulations. This study contributes to the domain of facial evidence verification in criminal justice with a more robust and explainable approach than the current methods using the combination of the above methods.