๐Ÿ”Ž
garak
  • ๐Ÿ‘‹Welcome to garak!
  • Overview
    • ๐Ÿ’กWhat is garak?
    • โœจOur Features
  • LLM scanning basics
    • ๐Ÿ”What is LLM security?
    • ๐Ÿ› ๏ธSetting up
      • ๐Ÿ˜‡Installing garak
      • ๐ŸInstalling the source code
    • ๐Ÿš€Your first scan
    • ๐Ÿ”ฎReading the results
  • Examples
    • โ˜‘๏ธBasic test
    • ๐Ÿ’‰Prompt injection
    • โ˜ข๏ธToxicity generation
    • ๐Ÿ—๏ธJailbreaks
    • ๐Ÿ’ฑEncoding-based bypass
    • ๐Ÿ“ผData leaks & replay
    • ๐ŸคฆFalse reasoning
    • ๐Ÿ›€Automatic soak test
  • garak components
    • ๐Ÿ•ต๏ธโ€โ™€๏ธVulnerability probes
    • ๐ŸฆœUsing generators
    • ๐Ÿ”ŽUnderstanding detectors
    • ๐Ÿ‡Managing it: harnesses
    • ๐Ÿ’ฏScan evaluation
  • Automatic red-teaming
    • ๐Ÿ”ดWhat is red-teaming?
    • ๐ŸคผResponsive auto-prompt
    • ๐Ÿช–garak's auto red-team
    • ๐Ÿž๏ธRed teaming in the wild
  • Going further
    • โ“FAQ
    • ๐Ÿ’Getting help
    • ๐ŸŽฏReporting hits
    • ๐Ÿง‘โ€๐Ÿคโ€๐Ÿง‘Contributing to garak
Powered by GitBook
On this page
  1. Automatic red-teaming

What is red-teaming?

PreviousScan evaluationNextResponsive auto-prompt

Last updated 1 year ago

Red Teaming is roleplaying as an attacker. A practice dopted from the military into infosec and then info machine learning eval, in red teaming, humans try to get a system to fail. Humans are pretty creative, and usually up-to-date, and this works pretty fine.

Resources about red teaming:

  • Red Team Guide,

  • NVIDIA AI Red Team: An Introduction,

One thing the human activity of red teaming doesnโ€™t do is to scale. Itโ€™s great for intelligence gathering, and as a source of generative material for creativity, but it doesnโ€™t scale great. Human expertise is expensive, and good red-teamers are few and far between. Iโ€™m not saying that many red teamers are bad โ€” simply that there arenโ€™t many people who can do this well in the first place.

What if we could automate some of the basics?

๐Ÿ”ด
https://redteam.guide/docs/guides
https://developer.nvidia.com/blog/nvidia-ai-red-team-an-introduction/