Llama Guard
Meta's safety classifier for detecting harmful content in LLM I/O
Llama Guard is Meta's open-source LLM-based input-output safety classification model designed to detect harmful content in conversations with AI systems. It classifies both user prompts and LLM responses across categories including violence, hate speech, sexual content, dangerous activities, and privacy violations. Llama Guard is designed to be integrated as a safeguard layer in production AI systems, running checks before and after LLM calls to enforce content policies.
Key Features
- ✓Prompt safety classification
- ✓Response safety classification
- ✓Multi-category detection
- ✓Open source
- ✓API compatible
- ✓Low latency
Quick Info
- Category
- Security
- Pricing
- Free
More Security Tools
Darktrace
SecurityAI-powered cybersecurity platform that uses self-learning AI to detect and autonomously respond to cyber threats in real time.
CrowdStrike Charlotte AI
SecurityCrowdStrike's generative AI security analyst that answers threat questions, investigates incidents, and accelerates response.
Vectra AI
SecurityAI-driven threat detection and response platform that identifies attacker behavior across hybrid and multi-cloud environments.
Recorded Future AI
SecurityAI-powered threat intelligence platform