The AI-authored Denial-of-Service attack
The article analyzes how LLM tools fundamentally alter the threat model for open source projects with strict AI-authored-code bans, creating a governance-based denial-of-service vector. By lowering the cost of generating code via AI while projects must perform expensive clean-room reimplementation, attackers gain asymmetric advantage—particularly via time-bomb attacks (merging non-trivial changes that are later revealed as AI-authored) or license violation arguments. Projects that selectively enforce AI policies face reputational and practical pressure, as reversing complex security fixes becomes costly or infeasible. The article argues that pragmatic policy enforcement paired with consequences remains more effective than absolutist bans that projects cannot sustain.