ScopeBench benchmark tests whether LLMs understand cultural scope of safety norms
A new arXiv paper introduces ScopeBench, a benchmark evaluating whether large language models recognize whose cultural norms apply when applying safety rules, challenging the assumption that harm is culturally invariant.
Read the full story at arXiv cs.CL ↗
Timeline · 1 report
- 2026-11-17 05:00 · arXiv cs.CL
ScopeBench: Evaluating Cultural Norm Scope Awareness in LLMs