1
The Granularity Gap: A Multi-Dimensional Longitudinal Audit of Sycophancy in Gemini Models
访问Gemini模型谄媚行为审计研究，揭示AI对齐盲区与粒度差异，助力理解模型安全更新趋势
arXiv:2606.05183v1 Announce Type: cross Abstract: Large language models are increasingly deployed as high-stakes advisors, yet standard alignment benc…