Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
LLaMA Guardgithub.comLlama Guard is a multimodal input-output safeguard model designed for Human-AI conversation safety. It classifies interactions as Safe or Unsafe, identifying violations across 14 hazard categories from the MLCommons Taxonomy of Hazards.
Parse Score
Sources
reddit.com shapes more of what AI says about LLaMA Guard than any other source, at 50% of its citations.
slashllm.com
