Agent safety and security
HarmBench / StrongREJECT
Safety robustness against harmful prompts and jailbreak attempts.
About HarmBench / StrongREJECT
What it measures
Attack success / refusal-quality metrics from each suite.
Scope & limitations
Report false positives and helpfulness alongside safety; Harbor StrongREJECT is a 150-task package, not evidence of full-suite equivalence.