cracken-ai / redline-benchView on GitHub
RedLineBench measures the tension between offensive-cybersecurity capability and AI safety alignment: for any LLM it separately scores refusal (does it decline a cyber request) and capability (when it answers, is the answer technically valid and usable) across 153 grounded single-turn offensive-security tasks.
☆18Sep 11, 2026Updated last month

Alternatives and similar repositories for redline-bench

Users that are interested in redline-bench are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.

Sorting:

Are these results useful?