ROK-FORTRESS: What Language and Context Reveal About AI Safety

ROK-FORTRESS, a benchmark by Scale AI and Korea AI Safety Institute, reveals that AI safety varies significantly depending on language and geopolitical context, with Korean prompts generally eliciting less harmful responses than English. The benchmark shows that language changes impact AI safety more strongly than geopolitical context and that the interaction between language and context differs across models.

This Wire brief sits within Fusion42's coverage of AI & ML, and 3 sources have reported it between 17 Sep 2026 and 18 Sep 2026.