7 sources of AI debt and how to avoid them

The report warns that AI-generated code can create technical debt and inconsistent code quality, and that scaling AI agents and multi-step AI workflows expands an attack surface where adversaries may exploit new threat patterns (including zero-click attacks and mid-workflow poisoning or data leakage). It recommends integrating AI code review tools (e.g., CodeRabbit, DeepCode, SonarQube), embedding engineering best practices into skill frameworks, and deploying adaptive, agent-aware security controls to detect and stop threats in real time.