1 paper
Zhihao Guo, Zonghan Wu, Huan Huo +6
Even well-aligned large language models confidently generate factually incorrect text, making hallucination a persistent reliability risk in high-stakes deployments. These models n…