AI Alignment Strategies from a Risk Perspective: Independent Safety Mechanisms or Shared Failures?
Ruhr-Universität Bochum(博尔塔伦大学);Rheinische Friedrich-Wilhelms-Universität Bonn(波恩莱茵-斐迪南-威廉大学);Lamarr Institute for Machine Learning and Artificial Intelligence(拉马尔人工智能与机器学习研究所)
AI alignment research aims to develop techniques to ensure that AI systems do not cause harm. However, every alignment technique has failure modes, which are conditions in which there is a non-negligible chance that the technique fails to provide safety. As a strategy for risk mitigation, the AI safety community has increasingly adopted a defense-in-depth framework: Conceding that there is no single technique which guarantees safety, defense-in-depth consists in having multiple redundant protections against safety failure, such that safety can be maintained even if some protections fail. However, the success of defense-in-depth depends on how (un)correlated failure modes are across alignment techniques. For example, if all techniques had the exact same failure modes, the defense-in-depth approach would provide no additional protection at all. In this paper, we analyze 7 representative alignment techniques and 7 failure modes to understand the extent to which they overlap. We then discuss our results' implications for understanding the current level of risk and how to prioritize AI alignment research in the future.
Chain of Risk: Safety Failures in Large Reasoning Models and Mitigation via Adaptive Multi-Principle Steering
Harvard University(哈佛大学);University of Southern California(南加州大学);Brown University(布朗大学);Pennsylvania State University(宾夕法尼亚州立大学);Texas A&M University(德克萨斯阿姆大学);Purdue University(普渡大学);University of California, Irvine(加州大学 Irvine 分校);Massachusetts Institute of Technology(麻省理工学院)
University of Hong Kong(香港大学);University of Texas at Austin(德克萨斯大学奥斯汀分校);Australian Catholic University(澳大利亚天主教大学);University of Southern California(南加州大学)
Moving the Safety Barrier: Dynamic Routing Adaptive Alignment Against White-Box Attacks
Nanjing University of Science and Technology(南京理工大学);The University of Sydney(悉尼大学);National University of Singapore(新加坡国立大学);University of New South Wales(新南威尔士大学);Nanjing University(南京大学)
When Models Outthink Their Safety: Unveiling and Mitigating Self-Jailbreak in Large Reasoning Models
Chinese Information Processing Laboratory, Institute of Software, Chinese Academy of Sciences(中国科学院软件研究所信息处理实验室);University of Chinese Academy of Sciences(中国科学院大学)
State-Dependent Safety Failures in Multi-Turn Language Model Interaction
School of Cyber Science and Technology, University of Science and Technology of China, China(中国科学技术大学信息科学与技术学院);Nanyang Technological University, Singapore(南洋理工大学);Tsinghua University, Beijing, China(清华大学);Beijing Electronic Science and Technology Institute, Beijing, China(北京电子科技研究所)
LongGuard: Mechanistic Analysis and Training-Free Mitigation of Long-Context Failure in Safety Guardrails
Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所);School of Cyber Security, University of Chinese Academy of Sciences(中国科学院大学网络空间安全学院)
StructBreak: Structural Cognitive Overload-Induced Safety Failures in MLLMs
Key Laboratory of Trustworthy Distributed Computing and Service (MoE), Beijing University of Posts and Telecommunications(可信分布式计算与服务重点实验室(MoE),北京邮电大学);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所)
Multilingual Safety Signals Are Multi-Layered: Filtering Safety-Degrading Data for Safer LLMs
Beihang University(北京航空航天大学);Zhongguancun Laboratory(中关村实验室);Hangzhou Innovation Institute, Beihang University(北京航空航天大学杭州创新研究院);Qingdao Research Institute, Beihang University(北京航空航天大学青岛研究院)
SafeRedirect: Defeating Internal Safety Collapse via Task-Completion Redirection in Frontier LLMs
Department of Computer Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, China(南方科技大学计算机科学与工程系,深圳518055,中国);The Hong Kong Polytechnic University, Hong Kong, China(香港理工大学,香港,中国);George Washington University(乔治华盛顿大学);School of Data Science, Lingnan University, Hong Kong, China(岭南大学数据科学学院,香港,中国)