Recent studies have revealed that LLMs can exhibit behavioral self-awareness: the ability to accurately describe or predict their own learned behaviors without explicit supervision. This capability raises safety concerns as it may, for example, allow models to better conceal their true abilities during evaluation. We attempt to characterize the minimal conditions under which such self-awareness emerges, and the mechanistic processes through which it manifests. Through controlled finetuning experiments on instruction-tuned LLMs with low-rank adapters (LoRA), we find: (1) that self-awareness can be reliably induced using a single rank-1 LoRA adapter; (2) that the learned self-aware behavior can be largely captured by a single steering vector in activation space, recovering nearly all of the fine-tune's behavioral effect; and (3) that self-awareness is non-universal and domain-localized, with independent representations across tasks. Together, these findings suggest that behavioral self-awareness emerges as a domain-specific, linear feature that can be easily induced and modulated.
The Reasoning Trap -- Logical Reasoning as a Mechanistic Pathway to Situational Awareness
MARS 4.0 Fellowship, Cambridge AI Safety Hub(CAISH), University of Cambridge(MARS 4.0 Fellow,剑桥人工智能安全中心(CAISH),剑桥大学);AWS Generative AI Innovation Center, Amazon Web Services, USA(亚马逊生成AI创新中心,亚马逊网络服务,美国);Google, USA(谷歌,美国);Stanford University(斯坦福大学);Northeastern University, Seattle, WA, USA(东北大学,西雅图,华盛顿州,美国)
Behind Harmful Compliance: Behavioral and Mechanistic Divergence Across LLM Jailbreaks
Department of Computer Science(计算机科学系);Luddy School of Informatics, Computing, and Engineering(信息学、计算与工程学院);Indiana University Bloomington(印第安纳大学布卢明顿分校)
ETH Zürich(苏黎世联邦理工学院);ELLIS Institute Tübingen(图宾根ELLIS研究所);Max Planck Institute for Intelligent Systems(智能系统马克斯·普朗克研究院);Tübingen AI Center(图宾根人工智能中心);University of Toronto & Vector Institute(多伦多大学及向量研究所);EuroSafeAI(欧洲安全人工智能)
Mechanistic Analysis of Catastrophic Forgetting in Large Language Models During Continual Fine-tuning
Division of Statistics and Machine Learning (STIMA), Department of Computer and Information Science (IDA), Linköping University(统计与机器学习系(STIMA)、计算机与信息科学系(IDA)、利厄普堡大学)
Agent MechSuits: Mechanistic Subspace Safety Steering for Multi-Turn CLI Agents
University of Georgia(佐治亚大学);University of South Florida(南佛罗里达大学);Rutgers University(罗格斯大学);University of Southern California(南加州大学);Johns Hopkins University(约翰霍普金斯大学)
Weight Patching: Toward Source-Level Mechanistic Localization in LLMs
Key Laboratory of Ministry of Education for Brain Inspired Intelligent Perception and Cognition, University of Science and Technology of China(教育部脑科学与智能感知认知重点实验室,中国科学技术大学)
MindScape Study: Integrating LLM and Behavioral Sensing for Personalized AI-Driven Journaling Experiences
Dartmouth College(达特茅斯学院);Colby College(科尔比学院);Brown University(布朗大学);Cornell Tech(康奈尔科技学院);Massachusetts Institute of Technology(麻省理工学院);University College London(伦敦大学学院);Biocogniv Inc(比奥认知公司);University of California, San Diego(加利福尼亚大学圣迭戈分校);Microsoft Research(微软研究院)
From Human Cognition to Neural Activations: Probing the Computational Primitives of Spatial Reasoning in LLMs
School of Information Science, Beijing Language and Culture University(北京语言大学信息科学学院);School of Artificial Intelligence, Beijing Normal University(北京师范大学人工智能学院)
Theory of Mind and Self-Attributions of Mentality are Dissociable in LLMs
Google, Paradigms of Intelligence Team(Google,智能范式团队);Knowledge Lab, University of Chicago(芝加哥大学知识实验室);Institute of Philosophy, School of Advanced Study, University of London(伦敦大学高等研究院哲学研究所);Department of Biomedical Informatics and Medical Education and Department of Bioethics and Humanities, School of Medicine, University of Washington(华盛顿大学医学院生物医学信息学与医学教育系及生物伦理学与人文学系);Kellogg School of Management, Northwestern University(西北大学凯洛格管理学院);Santa Fe Institute(圣塔菲研究所)
When Can Large Reasoning Models Save Thinking? Mechanistic Analysis of Behavioral Divergence in Reasoning
State Key Laboratory for Novel Software Technology, Nanjing University, China(新型软件技术国家重点实验室,南京大学,中国);University of California, Merced, USA(加州大学梅德福分校);National Institute of Healthcare Data Science, Nanjing University, China(健康医疗数据科学国立研究所,南京大学,中国)
Large language models replicate and predict human cooperation across experiments in game theory
Department for Computational Social Sciences and Humanities, Barcelona Supercomputing Center(巴塞罗那超级计算中心计算社会科学与人文学系);Grupo de Sistemas Complejos, Universidad Politécnica de Madrid(马德里理工大学复杂系统研究组);School of Law and Political Science, Universitat Oberta de Catalunya(加泰罗尼亚开放大学法律与政治科学学院)
Probing Cultural Awareness in LLMs: A Case Study of Cross-Culture Aesthetic Stylistics
Department of Computing, Hong Kong Polytechnic University(香港理工大学计算机系);Institute of Digital Twin, Eastern Institute of Technology, Ningbo(宁波东部技术研究所数字孪生研究所);Department of Language Science and Technology, Hong Kong Polytechnic University(香港理工大学语言科学与技术系);School of Design, Hong Kong Polytechnic University(香港理工大学设计学院);Research Institute for Quantum Technology, Hong Kong Polytechnic University(香港理工大学量子技术研究所);Department of Communication Science, Vrije Universiteit Amsterdam(阿姆斯特丹自由大学传播科学系)
BioLLMAgent: A Hybrid Framework with Enhanced Structural Interpretability for Simulating Human Decision-Making in Computational Psychiatry
School of Computer Science and Technology, East China Normal University(东华大学计算机科学与技术学院);Department of Computer Science, University of Reading(Reading大学计算机科学系);Department of Computer Science, Brunel University of London(伦敦布鲁内尔大学计算机科学系)