Atlas · GenAI 2026
OWASP Top 10 for LLM Applications
The canonical risk checklist for LLM apps — prompt injection, insecure output handling, data poisoning, excessive agency.
conceptPeak: 2024Security FrameworksAI consensus: 0/3
Prerequisites
- mediumPrompt Injection Defense
Prompt injection is the #1 item on the list.
Recommended reference
OWASP Top 10 for Large Language Model Applications — OWASP, 2025