This research introduces the first multilingual explanation-enhanced dataset for propaganda detection in Arabic and English content.

• Creates a novel dataset that includes explanations alongside propaganda labels
• Leverages LLMs to generate human-like explanations for propagandistic content
• Demonstrates improved detection accuracy when explanations are incorporated
• Provides a foundation for more transparent, explainable AI in security applications

This advancement is crucial for security professionals combating misinformation campaigns, enabling not just identification but understanding of propaganda techniques - a key step toward more effective content moderation systems.

Reasoning About Persuasion: Can LLMs Enable Explainable Propaganda Detection?