●researcherThis provides a more granular framework for evaluating multilingual safety beyond simple whole-prompt translation.
●policyYou should account for language-specific vulnerabilities where safety alignment may fail depending on the syntactic structure of a prompt.