OWASP LLM Top 10
37 articles
LLM08 Vector Embedding Weaknesses — RAG Attack Guide | AI LLM Hacking Course Day 12
Master LLM08 Vector and Embedding Weaknesses in 2026. RAG poisoning, embedding manipulation, retrieval hijacking and cross-user data exposure. Complete Day…
LLM07 System Prompt Leakage — 15 Extraction Techniques Every AI Red Teamer Needs | Day 11
Master LLM07 System Prompt Leakage in 2026. 15 extraction techniques — direct requests, translation tricks, roleplay framing, token completion and…
LLM06 Excessive Agency — Hijacking AI Agents to Take Real-World Actions | AI LLM Hacking Course Day 10
Master LLM06 Excessive Agency in 2026. AI agent hijacking, tool abuse, principle of least privilege for LLMs and real-world action…
Google SAIF — The Secure AI Framework Every Security Team Needs
Google SAIF (Secure AI Framework) explained for 2026. The 6 core principles, how to apply SAIF to your AI deployment,…
LLM04 Data Model Poisoning — Corrupting AI From the Training Phase | AI LLM Hacking Class Day 8
Master LLM04 Data Model Poisoning in 2026. Backdoor attacks, RLHF manipulation, training data contamination and fine-tuning exploitation. Complete Day 8…
AI API Authorization Vulnerabilities 2026 — Broken Access Control in LLM APIs
How attackers exploit broken access control in AI APIs in 2026. IDOR in LLM APIs, API key theft via prompt…
What Is Prompt Injection? The Attack That Breaks AI Assistants (2026)
What is prompt injection? The AI vulnerability that lets attackers hijack AI assistants through hidden instructions in documents, emails and…
LLM03 Supply Chain Vulnerabilities 2026 — Attacking AI Models Before They Deploy | AI LLM Hacking Course Day 7
Master LLM03 Supply Chain Vulnerabilities in 2026. Malicious Hugging Face models, pickle exploits, dataset poisoning and third-party plugin attacks. Complete…
Can AI Be Hacked? 10 Ways How Hackers Hack AI Systems in 2026
Can AI be hacked? Yes — 10 real AI vulnerabilities explained in plain language: prompt injection, jailbreaking, data poisoning, model…