NVIDIA/garak — The LLM vulnerability scanner

Description and use

NVIDIA/garak is an open‑source vulnerability scanner designed specifically for large language models (LLMs). It provides automated detection of security flaws, bias issues, and prompt‑injection weaknesses across a variety of LLM architectures, helping developers and researchers harden their AI systems. The tool integrates seamlessly into AI/ML pipelines and offers extensible test suites for continuous security assessment.

Problem

NVIDIA/garak helps add AI capability without starting from a blank model stack.

Outcome

Medium: useful for prototypes or internal workflows, with dependency and maintenance checks before production.

Workflow

• prototype AI features
• build RAG or agent workflows
• evaluate model tooling

Evidence

the LLM vulnerability scanner

Source