https://uncensoredainews.com/news/new-study-reveals-why-safety-aligned-ai-still-over-refuses-harmless-queries-25752026-10-31T22:28:43ZUncensored AI Newsen2026-10-31T22:28:43ZPreprint Explains Why Safety-Aligned LLMs Over-Refuse Harmless Querieshttps://uncensoredainews.com/news/new-msafetemp-dataset-targets-multilingual-ai-safety-gaps-25692026-10-30T22:27:41ZUncensored AI Newsen2026-10-30T22:27:41ZMSafeTemp Preprint Offers Multilingual Dataset for AI Safety Research