Who is Creating Malware Repositories on GitHub and Why?
Nishat Ara Tania, Md Rayhanul Masud, Md Omar Faruk Rokon, Qian Zhang, Michalis Faloutsos · 2024
Recent studies have found thousands of malware source code repositories on GitHub. For the first time, we propose to understand the origins and motivations behind the creation of such malware repositories. For that, we collect and profile the authors of malware repositories using a three-fold systematic approach. First, we identify 14K users in GitHub who have authored at least one malware repository. Second, we leverage a pretrained large language model (LLM) to estimate the likelihood of malicious intent of these authors. This innovative approach led us to categorize 3339 as Malicious, 3354 as Likely Malicious, and 7574 as Benign authors. Further, to validate the accuracy and reliability of our classification, we conduct a manual review of 200 randomly selected authors. Third, our analysis provides insights into the authors' profiles and motivations. We find that Malicious authors often have sparse profiles and focus on creating and spreading malware, while Benign authors typically have complete profiles with a focus on cybersecurity research and education. Likely Malicious authors show varying levels of engagement and ambiguous intentions. We see our study as a key step towards understanding the ecosystem of malware authorship on GitHub.