一作ICML 2026Poster· Social Aspects / Alignment
Alignment Tampering: How Reinforcement Learning from Human Feedback Is Exploited to Optimize Misaligned Biases
Dongyoon Hahm · Dylan Hadfield-Menell · Kimin Lee
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。
Dongyoon Hahm · Dylan Hadfield-Menell · Kimin Lee
Suchae Jeong · Jaehwi Song · Haeone Lee · Hanna Kim · Jian Kim · Dongjun Lee · Dong Kyu Shin · Changyeon Kim · Dongyoon Hahm · Woogyeol Jin · Juheon Choi · Kimin Lee