AI 安全研究者
Anthropic
The Capacity for Moral Self-Correction in Large Language Models
Toy Models of Superposition
Scaling Laws and Interpretability of Learning from Repeated Data
Predictability and Surprise in Large Generative Models
A General Language Assistant as a Laboratory for Alignment
署名以 dblp 和本人的 Google Scholar 为准；只列本站收录的 AI 安全论文。