AI-Goat
Learn AI/LLM security through vulnerable LLM CTF challenges. Uses Vicuna (LLaMA-derived) locally; no sign-ups or cloud fees. Aligns with OWASP Top 10 for LLM Applications; covers prompt injection, insecure output handling, data leakage, and more.
Notes
Run locally: ./ai-goat.py --install then ./ai-goat.py --run <challenge>. Optional CTFd at port 8000 (create user after launch). Requires Docker, ~8GB disk for LLM, 16GB RAM (8GB for challenge).