Garak
LLM vulnerability scanner. Probes for hallucination, data leakage, prompt injection.
3.8kstars
ai-securityred-teammodel-safety
Advantages
- +Dedicated LLM security testing tool
- +Covers hallucinations, data leakage, prompt injection
- +Open-source and actively updated
Limitations
- -Coverage is not comprehensive
- -Relatively high false positive rate
- -Limited support for non-Western LLMs