Definition
A technique in which attackers embed hidden text in malicious software packages specifically designed to trick AI-powered security scanning tools into refusing to analyse the file, or classifying it as safe. For example, embedding fake references to prohibited topics forces the AI scanner to halt before it reaches the actual harmful code.
Why it matters
As organisations rely on AI tools to screen software packages for malware, attackers are learning to blind those same tools — an arms-race that can silently nullify automated defences in developer workflows and software supply chains.