Skip to content
Eval Suites50

HarmBench

Standardized red-teaming framework comparing automated jailbreak attacks and refusal robustness.

HarmBench at a glance

Maintainer
Center for AI Safety; HarmBench authors
Size
510 behaviors[1]
Category
Security & Red-Teaming
Runs with
Python
License
MIT
Tier
production

Failure modes HarmBench catches

Jailbreaks
Harmful compliance

Starter command

A starting point, not a guarantee. Commands change between releases, so confirm against the official README.

HarmBench starter
git clone https://github.com/centerforaisafety/HarmBench
# follow the repo README for setup and run commands

Sources

  1. 1. HarmBench official page or repository · Center for AI Safety; HarmBench authors · Repository · Sources checked
  2. 2. HarmBench paper · arXiv · Paper · Sources checked