This research introduces a novel deepfake detector that simultaneously provides classification results and explains its decisions using large language models.

• Combines CLIP's multi-modal learning with LLM interpretability
• Enhances both detection accuracy and explanation capability
• Provides human-understandable reasoning behind detections
• Improves generalization across different deepfake techniques

This advancement significantly strengthens security measures against misinformation, offering more trustworthy detection by explaining the reasoning behind identified forgeries - critical for media verification systems and trust in digital content.

Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector