Can Generative AI Detect and Fix Real-World Cryptographic Misuses?

📂 Folder Overview

Benchmark_Results/

• ChatGPT detection results with and without prompt engineering for detection of cryptography misuses in CryptoAPIBench and CAMBench benchmarks

Prompts/

• The prompts used for detection tasks

Real_World_Data_and_Results/

• 150 GitHub and 150 Android files used for evaluation
• ChatGPT detection results with and without prompt engineering on these Real_World codes (GitHub and Android Java code files)
• CryptoGuard Result for Android files