This research introduces IPAD (Inverse Prompt for AI Detection), a novel framework that addresses critical limitations in current AI text detectors.

• Creates inverse prompts to predict what instructions might have generated the text
• Achieves superior robustness against out-of-distribution data and common evasion attacks
• Provides explainable evidence by revealing potential prompts behind AI-generated content
• Demonstrates significantly better performance than existing detectors on texts perturbed by paraphrasing or editing

This advancement is crucial for security applications as it helps combat potential misuse of AI systems while providing transparent reasoning for its classifications—essential for regulatory compliance and trust in detection systems.

IPAD: Inverse Prompt for AI Detection -- A Robust and Explainable LLM-Generated Text Detector