Linux ELF Malware Generator Evades ML Detection With Semantic-Preserving Changes

Researchers have developed a Linux ELF malware generator that uses semantic-preserving transformations to evade machine learning detection with a 67.74% evasion rate. The generator’s success highlights a vulnerability in ML models' sensitivity to benign strings within executables, indicating a need for advanced detection methods beyond simple string analysis.