LLM judge prompt variations alone shift HarmBench harmful-response rates by up to 24.2 percentage points and produce moderate instability in model safety rankings.
arXiv preprint arXiv:2603.14723 (2026)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
How Sensitive Are Safety Benchmarks to Judge Configuration Choices?
LLM judge prompt variations alone shift HarmBench harmful-response rates by up to 24.2 percentage points and produce moderate instability in model safety rankings.