# Garak — LLM Vulnerability Scanner by NVIDIA

> Automated red-teaming for LLMs — finds vulnerabilities before attackers do.

- Org: NVIDIA
- Date: 2024-12-15
- Category: security, tool, repo
- Source: https://github.com/NVIDIA/garak
- Canonical page: https://ai-tldr.dev/releases/nvidia-garak/

## In short

NVIDIA's open-source LLM vulnerability scanner. Probes for prompt injection, data leakage, hallucination, toxicity, and jailbreaks across 50+ attack families.

## What it is

Garak is NVIDIA's open-source tool for probing LLM vulnerabilities. It tests models against prompt injection, jailbreaks, data leakage, and harmful content generation using 50+ attack families.

## How it works

You point Garak at any LLM endpoint (OpenAI, local models, HuggingFace). It runs thousands of adversarial prompts organized by attack type, logs which ones succeed, and generates a vulnerability report.

## Why it matters

Before deploying an LLM, you need to know its weaknesses. Garak automates what would take a red team weeks — systematic probing across known attack vectors.

## Who it's for

Security teams, ML engineers deploying LLMs to production.

## Try it

pip install garak && garak --model_type openai --model_name gpt-4

## Links

- [GitHub](https://github.com/NVIDIA/garak)
- [Docs](https://docs.garak.ai)