作者位次 #2/4ACL 2026Findings· Findings of ACL
Thinking Twice Makes Large Language Models Safer and More Helpful
Yutao Mou · Yuxiao Luo · Shikun Zhang · Wei Ye
摘要
Current safety alignment techniques for large language models (LLMs) struggle to balance harmlessness and helpfulness: improving safety often comes at the cost of degraded utility. Our preliminary study shows that guiding unaligned base models with safety-aware reasoning that includes explicit se...