The Cama framework systematically evaluates how effectively code-specialized LLMs can identify and analyze Android malware, addressing unique challenges in decompiled Android code.

• Tackles the challenge of large function volumes in Android applications
• Evaluates LLM performance in identifying malicious functions without meaningful names
• Provides a structured approach to benchmark LLM capabilities for malware detection
• Bridges the gap between code intelligence and security applications

This research matters for security teams by establishing how LLMs can be leveraged as practical tools in mobile security operations, potentially improving detection capabilities for increasingly sophisticated Android malware threats.

On Benchmarking Code LLMs for Android Malware Analysis