OWASP LLM Top 10: From Web Roots to AI Frontiers

A 22-episode series on LLM application security, written from the infrastructure side: someone who has run red team engagements, written RBAC policies at scale, and debugged IAM privilege escalation — now applying that operational knowledge to a probabilistic application layer.

Who it’s for: Senior DevOps engineers, SREs, and security engineers who know Kubernetes, AWS/GCP/Azure IAM, and classic OWASP, but haven’t worked in LLM security yet.

Format: Foundation episodes (Part I) build the mental model. Vulnerability deep dives (Parts II–III) use Red/Detect/Defend structure — attack anatomy, what your SIEM catches and misses, and the defense-in-depth fix that actually holds.

This series is written thematically, not strictly in episode order — a high-relevance category (Excessive Agency, EP10) shipped ahead of ones still in progress. Each post stands alone; the breadcrumb tells you what it builds on.

Published Episodes

EP	Title
EP01	OWASP Top 10 History: How the List Evolved from 2003 to 2025
EP02	The Four OWASP Lists: Web App, API, Cloud-Native, and LLM Compared
EP03	Why Classic OWASP Breaks Down for LLMs: The New Attack Surface
EP04	OWASP LLM Top 10 2025: The Complete Map for DevSecOps
EP05	Prompt Injection Attacks: How LLM01 Becomes Full System Compromise
EP06	LLM Sensitive Information Disclosure: When the Model Becomes the Data Leak
EP10	LLM Excessive Agency: When Your AI Agent Goes Off-Script

Part I is complete. EP05 and EP06 open Part II — the vulnerability deep dives.

Coming Up

The rest of Part II (LLM03 Supply Chain through LLM10 Unbounded Consumption), Part IV (red team tools, runtime defense, compliance frameworks, audits, roles), and Part V (career path into AI security) — 15 more episodes planned.

Start with EP01: OWASP Top 10 History →

Get new episodes in your inbox → subscribe