LAMD is a novel framework that leverages large language models to detect and classify Android malware by focusing on security-critical code regions, overcoming limitations of traditional approaches.

• Addresses key challenges by extracting and contextualizing relevant code sections rather than analyzing entire applications
• Achieves superior detection capabilities through zero-shot inference without requiring extensive training data
• Provides human-readable explanations for its security decisions, enhancing transparency
• Demonstrates resilience against evolving malware techniques that often evade conventional detection systems

This research marks a significant advancement for mobile security teams by offering more adaptable and explainable malware detection, crucial for protecting against rapidly evolving threats in the Android ecosystem.

LAMD: Context-driven Android Malware Detection and Classification with LLMs