9 models + 1 dataset for detecting cultural hallucinations in Arabic LLM outputs. ICAIRE 2026 Track 3.