Threat group TeamPCP employs machine learning to accelerate attacks against open-source software repositories

Researchers from Oligo Security disclosed that TeamPCP, a cybercriminal group, operated continuously from 2020 onwards and compromised over 1,000 open-source software repositories. In recent months, the group integrated machine learning capabilities into its operations to automatically adjust malware code against diverse environments. The investigation revealed the creation of a novel worm capable of autonomous propagation across hijacked artificial intelligence systems. The application of AI technologies facilitated both faster iteration of attack methods and larger-scale coordinated campaigns, taking advantage of vulnerabilities introduced by organizations' increased reliance on machine learning infrastructure and freely-available code repositories.