一作ICLR 2026Poster· Social Aspects / Fairness, Equity, Justice and Safety
Self-Destructive Language Models
Yuhui Wang · Rongyi Zhu · Ting Wang
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Yuhui Wang · Rongyi Zhu · Ting Wang
Yuhui Wang · Changjiang Li · Guangke Chen · Jiacheng Liang · Ting Wang
Jiacheng Liang · Tanqiu Jiang · Yuhui Wang · Rongyi Zhu · Fenglong Ma · Ting Wang
This paper presents AutoRAN, the first framework to automate the hijacking of internal safety reasoning in large reasoning models (LRMs). At its core, AutoRAN pioneers an execution simulation paradigm that leverages a weaker but less-aligned model to simulate execution reasoning for initial hijac...
Yuhui Wang · Tanqiu Jiang · Jiacheng Liang · Charles Fleming · Ting Wang