ICLR

International Conference on Learning Representations · 会议 · Machine Learning

2026-08-14 至 2026-08-14 收录 1 篇

2602.06854 2026-08-14 cs.CL 版本更新

SEMA: Simple yet Effective Learning for Multi-Turn Jailbreak Attacks

SEMA: 简单却有效的多轮对抗攻击学习

Mingqian Feng, Xiaodong Liu, Weiwei Yang, Jialin Song, Xuekai Zhu, Chenliang Xu, Jianfeng Gao

AI总结 SEMA通过多阶段学习实现高效多轮对抗攻击，无需依赖现有策略，提升攻击成功率并增强安全测试能力。

Comments ICLR 2026, 37 pages, 13 tables, 7 figures

↑