Benchmark重罪基准
Felony Bench
A red-team benchmark dataset designed specifically to evaluate offensive capabilities of large language models. Test items cover criminal scenarios including network intrusion, malicious code generation, and social engineering attacks, assessing a model's ability to autonomously execute complete attack chains.
Timeline (last 90 days)
Sep 19
Felony Bench is a red-teaming benchmark specifically designed to evaluate large models' offensive capabilities in illegal or high-risk behavioral scenarios, covering scenarios such as network intrusion, malicious code generation, and social engineering attacks.
Unverified50%