AI Governance Tools Field reference
Chapter 03 · Attack

AI red teaming tools

Short answer · ai red teaming tools

This chapter lists 3 evidence-gated tools (garak (NVIDIA), Mindgard, Promptfoo), backed by 108 single-source dated facts that cleared deterministic and two-judge review.

§3.1 Listing

Red teaming tools by licence

Licence unknown

Table 3.1cRed teaming tools with unverified licence
#ToolLicenceWhat it doesEvidenceLast fact
01garak (NVIDIA)Licence unknownAppSecSanta describes garak as NVIDIA’s open-source LLM vulnerability scanner with 50+ probe modules for prompt injection, jailbreaks, and hallucinations.
02MindgardLicence unknownMindgard's GitHub profile describes it as a provider of AI red teaming and security testing solutions.
03PromptfooLicence unknownTechCrunch describes Promptfoo as an AI security startup founded in 2024 to protect LLMs from online adversaries.
§3.2 Evidence

Gated evidence highlights

garak (NVIDIA)

AppSecSanta describes garak as NVIDIA’s open-source LLM vulnerability scanner with 50+ probe modules for prompt injection, jailbreaks, and hallucinations.

  • AppSecSanta describes garak as NVIDIA’s open-source LLM vulnerability scanner with 50+ probe modules for prompt injection, jailbreaks, and hallucinations. (appsecsanta.com, verified )
  • NHIMG says NVIDIA officially contributes to garak. (nhimg.org, verified )
  • garak’s website says NVIDIA and the community actively update the open-source software. (garak.ai, verified )
  • garak’s website says the research-led project is backed by NVIDIA. (garak.ai, verified )
  • The garak paper on arXiv describes a Python framework distributed under the Apache 2.0 license for finding vulnerabilities in LLM-based technologies, systems, applications, and services. (arxiv.org, verified )
  • The garak paper on arXiv says native integrations include models from Hugging Face, Cohere, OpenAI, and NVIDIA NIMs, along with gguf models, Replicate and Octo ML platforms, Python functions, and a REST connector. (arxiv.org, verified )

Mindgard

Mindgard's GitHub profile describes it as a provider of AI red teaming and security testing solutions.

  • Mindgard's GitHub profile describes it as a provider of AI red teaming and security testing solutions. (github.com, verified )
  • SecurityWeek says Mindgard was founded in 2022 as a Lancaster University spinout and is headquartered in London and Boston. (securityweek.com, verified )
  • The BBC identifies Peter Garraghan as Mindgard's founder. (bbc.com, verified )
  • Business Wire describes Mindgard as providing AI security solutions to help enterprises discover and assess their AI systems and defend them. (businesswire.com, verified )
  • Mindgard says its integrations with Burp Suite and ServiceNow, as well as Splunk, connect AI security testing to existing development and security workflows. (mindgard.ai, verified )
  • Mindgard says deployment options include APIs, CI/CD pipelines, Burp Suite, and a single-click workflow. (mindgard.io, verified )

Promptfoo

TechCrunch describes Promptfoo as an AI security startup founded in 2024 to protect LLMs from online adversaries.

  • TechCrunch describes Promptfoo as an AI security startup founded in 2024 to protect LLMs from online adversaries. (techcrunch.com, verified )
  • CNBC describes Promptfoo as a cybersecurity startup providing tools to safeguard and test complex artificial intelligence systems. (cnbc.com, verified )
  • OpenAI describes Promptfoo as an AI security platform that helps enterprises identify and remediate vulnerabilities during development. (openai.com, verified )
  • TechCrunch describes Promptfoo's LLM security-testing tools as including an open-source interface and library. (techcrunch.com, verified )
  • CNBC says Promptfoo's open-source project lets developers test AI prompts and agents and compare large language model performance. (cnbc.com, verified )
  • In 2025, Georgetown reported that Promptfoo has a graphical user interface for red-teaming and data analysis. (cset.georgetown.edu, verified )
§3.3 Citations

Sources

  1. app.dealroom.co/news/note/mindgard-raises-30m-series-a-to-red-team-ai-systems, accessed
  2. github.com/mindgard, accessed
  3. mindgard.io, accessed
  4. mindgard.io/services/ai-red-teaming.html, accessed
  5. pulse2.com/mindgard-8-million-raised-for-safeguarding-ai-with-security-solutions, accessed
  6. appsecsanta.com/garak, accessed
  7. arxiv.org/html/2406.11036v1, accessed
  8. cset.georgetown.edu/article/ai-red-teaming-design-threat-models-and-tools, accessed
  9. discuss.huggingface.co/t/looking-for-services-to-validate-user-queries-for-content-and-security/171801, accessed
  10. docs.garak.ai/garak, accessed
  11. garak.ai, accessed
  12. github.com/NVIDIA/garak, accessed
  13. github.com/promptfoo/promptfoo/blob/9109019ac2d997dbd86f528421bef6da310aaae1/site/docs/configuration/expected-outputs/guardrails.md, accessed
  14. github.com/promptfoo/promptfoo/blob/fd8676e07c9827dd41a0a37b4a3dc538b186821e/site/docs/configuration/expected-outputs/guardrails.md, accessed
  15. github.com/promptfoo/promptfoo/blob/main/site/docs/guides/testing-guardrails.md, accessed
  16. github.com/promptfoo/promptfoo/security, accessed
  17. mindgard.ai, accessed
  18. mindgard.ai/automated-ai-red-teaming, accessed
  19. mindgard.ai/learn/customers, accessed
  20. mindgard.ai/legal/security-overview, accessed
  21. nhimg.org/garak-guide, accessed
  22. openai.com/index/openai-to-acquire-promptfoo, accessed
  23. techcrunch.com/2026/03/09/openai-acquires-promptfoo-to-secure-its-ai-agents, accessed
  24. bbc.com/news/articles/cmrergq3j7lgo, accessed
  25. businesswire.com/news/home/20260528871533/en/Mindgard-Launches-GuardBuster-to-Measure-How-AI-Guardrails-Perform-in-Real-World-Environments, accessed
  26. cnbc.com/2026/03/09/open-ai-cybersecurity-promptfoo-ai-agents.html, accessed
  27. databricks.com/blog/ai-security-action-applying-nvidias-garak-llms-databricks, accessed
  28. helpnetsecurity.com/2025/09/10/garak-open-source-llm-vulnerability-scanner, accessed
  29. promptfoo.dev, accessed
  30. promptfoo.dev/docs/intro, accessed
  31. promptfoo.dev/pricing, accessed
  32. promptfoo.dev/red-teaming, accessed
  33. securityweek.com/mindgard-raises-30-million-to-protect-ai-systems, accessed