Xiaohao Cai
1 indexed paper
Recent (6 mo)
1With code
0Influential cites
0Benchmarked
0Publications per year
126
Top categories
ML×1Crypto×1
Frequent co-authors
Research Timeline
2026
When Autoregressive Consistency Hurts Safety Alignment
The paper argues that shallow safety alignment in LLMs is due to autoregressive consistency, a mechanism that allows small harmful inputs to redirect the model's generation to unsafe outputs, necessitating adversarial safety training.
Highlighted terms show continued research focus across papers