LLM Safety

LLM: Large Language Model

Follow this topic Ask about this topic

Momentum

36 papers in the last four weeks, up 200% on the four weeks before. 0.4% of all new papers.

Jul 13Week of Sep 28

Latest papers 272

All topics

Chain of Risk: Safety Failures in Large Reasoning Models and Mitigation via Adaptive Multi-Principle Steering
May 7, 2026Xiaomin Li, Jianheng Hou, Zheyuan Deng +6LLM Safety LLM Safety Evaluation
Harvard University · University of Southern California · Brown University +6
One Turn Too Late: Learning When to Intervene Against Multi-Turn Malicious Intent
May 7, 2026Xinjie Shen, Rongzhe Wei, Peizhi Niu +6Language Model Safety Evaluation LLM Guardrails
Georgia Institute of Technology · University of Illinois Urbana-Champaign · UCSD +3
The Refusal--Compliance Tradeoff: A Large-Scale Safety Behavior Audit of Large Language Models
May 6, 2026Alif Al Hasan, Sumon BiswasLLM Safety Benchmarks Language Model Safety Evaluation
Department of Computer and Data Sciences Case Western Reserve University Cleveland, OH, USA
Understanding Annotator Safety Policy with Interpretability
May 6, 2026Alex Oesterling, Donghao Ren, Yannick Assogba +4Annotator Disagreement Human-in-the-Loop Annotation
Harvard University, Cambridge, MA, USA · Apple, Seattle, WA, USA · Apple, Cambridge, MA, USA +1
From Parameter Dynamics to Risk Scoring : Quantifying Sample-Level Safety Degradation in LLM Fine-tuning
May 6, 2026Xiao Wang, Yifei Zhang, YongKang Liu +4LLM Fine-Tuning LLM Safety
School of Computer Science and Engineering, Northeastern University, Shenyang, China · School of Computer and Communication Engineering, Northeastern University, Qinhuangdao, China
Reducing Conversational Escalation in Large Language Model Dialogue with Nonviolent Communication Constraints
May 1, 2026Zhixing Sun, Shenghe Xu, Tao LiLLM Prompting LLM Safety
Beijing University of Posts and Telecommunications Beijing, China · City University of Hong Kong Hong Kong SAR, China
FinSafetyBench: Evaluating LLM Safety in Real-World Financial Scenarios
May 1, 2026Yutao Hou, Yihan Jiang, Yuhan Xie +5Multilingual Language Model Evaluation Language Model Safety Evaluation
Shanghai University of Finance and Economics · Beihang University · Southern University of Science and Technology +1
Minimal, Local, Causal Explanations for Jailbreak Success in Large Language Models
Apr 30, 2026Shubham Kumar, Narendra AhujaLLM Interpretability Causal Reasoning in Language Models
University of Illinois Urbana-Champaign
Exploration Hacking: Can LLMs Learn to Resist RL Training?
Apr 30, 2026Eyon Jang, Damon Falck, Joschka Braun +6RL Exploration LLM Safety
MATS · UC San Diego · Anthropic +1
Perturbation Probing: A Two-Pass-per-Prompt Diagnostic for FFN Behavioral Circuits in Aligned LLMs
Apr 30, 2026Hongliang Liu, Tung-Ling Li, Yuhao WuLLM Interpretability Mechanistic Interpretability
Palo Alto Networks
Useless but Safe? Benchmarking Utility Recovery with User Intent Clarification in Multi-Turn Conversations
Apr 29, 2026Mingqian Zheng, Malia Morgan, Liwei Jiang +2LLM Safety Benchmarks Language Model Safety Evaluation
Carnegie Mellon University · Allen Institute for AI · University of Washington
Dynamic Adversarial Fine-Tuning Reorganizes Refusal Geometry
Apr 29, 2026Wenhao Lan, Shan Li, Xinhua Lai +4Adversarial Training Language Model Safety Evaluation
University of Chinese Academy of Sciences, Beijing, China · Inner Mongolia University of Technology, Inner Mongolia, China · Tsinghua University, Beijing, China +1
Culturally Aware GenAI Risks for Youth: Perspectives from Youth, Parents, and Teachers in a Non-Western Context
Apr 29, 2026Aljawharah Alzahrani, Tory Park, Tanusree SharmaLLM Safety Human-Centered AI
Pennsylvania State University
From Prompt Risk to Response Risk: Paired Analysis of Safety Behavior of Large Language Models
Apr 28, 2026Mengya Hu, Qiong Wei, Sandeep AtluriLanguage Model Safety Evaluation LLM Safety
Responsible AI, Microsoft
Below-Chance Blindness: Prompted Underperformance in Small LLMs Produces Positional Bias Rather than Answer Avoidance
Apr 28, 2026Jon-Paul CacioliLanguage Model Safety Evaluation LLM Safety
Independent Researcher, Melbourne, Australia
Jailbreaking Frontier Foundation Models Through Intention Deception
Apr 27, 2026Xinhe Wang, Katia Sycara, Yaqi XieJailbreak Attacks LLM Safety
Carnegie Mellon University
How Large Language Models Balance Internal Knowledge with User and Document Assertions
Apr 24, 2026Shuowei Li, Haoxin Li, Wenda Chu +1LLM Sycophancy Knowledge Augmentation for Language Models
Santa Clara University · Nanyang Technological University · California Institute of Technology
Emergent Strategic Reasoning Risks in AI: A Taxonomy-Driven Evaluation Framework
Apr 23, 2026Tharindu Kumarage, Lisa Bauer, Yao Ma +7Reward Hacking Language Model Safety Evaluation
Amazon Nova Responsible AI
Removing Sandbagging in LLMs by Training with Weak Supervision
Apr 23, 2026Emil Ryd, Henning Bartsch, Julian Stastny +2LLM Safety
MATS · University of Oxford · Redwood Research +1
Detoxification for LLM: From Dataset Itself
Apr 21, 2026Wei Shao, Yihang Wang, Gaoyu Zhu +4Training Data Curation LLM Safety
State Key Laboratory of AI Safety · Institute of Computing Technology, Chinese Academy of Sciences · University of Chinese Academy of Sciences
LLM Safety From Within: Detecting Harmful Content with Internal Representations
Apr 20, 2026Difan Jiao, Yilun Liu, Ye Yuan +4Language Model Safety Evaluation LLM Guardrails
University of Toronto · LMU Munich · McGill University
Different Paths to Harmful Compliance: Behavioral Side Effects and Mechanistic Divergence Across LLM Jailbreaks
Apr 20, 2026Md Rysul Kabir, Zoran TiganjLanguage Model Safety Evaluation LLM Interpretability
Department of Computer Science Luddy School of Informatics, Computing, and Engineering Indiana University Bloomington
Reverse Constitutional AI: A Framework for Controllable Toxic Data Generation via Probability-Clamped RLAIF
Apr 20, 2026Yuan Fang, Yiming Luo, Aimin Zhou +1Language Model Safety Evaluation Adversarial Scenario Generation
East China Normal University, Shanghai, China · Shanghai Innovation Institute, Shanghai, China
RoTRAG: Rule of Thumb Reasoning for Conversation Harm Detection with Retrieval-Augmented Generation
Apr 19, 2026Juhyeon Lee, Wonduk Seo, Junseo Koh +3Retrieval-Augmented Generation Toxicity Detection
Peking University · Enhans; Peking University · Enhans +1
Please refuse to answer me! Mitigating Over-Refusal in Large Language Models via Adaptive Contrastive Decoding
Apr 18, 2026Yupeng Qi, Ziyu Lyu, Lixin Cui +2Language Model Safety Evaluation LLM Refusal Behavior
School of Cyber Science and Technology, Shenzhen Campus of Sun Yat-sen University · School of Information, Central University of Finance and Economics · School of Artificial Intelligence, Beijing Normal University +1
DART: Mitigating Harm Drift in Difference-Aware LLMs via Distill-Audit-Repair Training
Apr 18, 2026Ziwen Pan, Zihan Liang, Jad Kabbara +1LLM Auditing Social Bias in Language Models
Emory University · MIT
Benign Fine-Tuning Breaks Safety Alignment in Audio LLMs
Apr 17, 2026Jaechul Roh, Virat Shejwalkar, Amir HoumansadrLanguage Model Safety Evaluation LLM Fine-Tuning
University of Massachusetts Amherst
SafeLM: Unified Privacy-Aware Optimization for Trustworthy Federated Large Language Models
Apr 17, 2026Noor Islam S. Mohammad, Uluğ BayazıtLLM Hallucination Mitigation Privacy-Preserving Language Models
Istanbul Technical University
A Systematic Study of Training-Free Methods for Trustworthy Large Language Models
Apr 17, 2026Wai Man Si, Mingjie Li, Michael Backes +1LLM Alignment LLM Safety
CISPA Helmholtz Center for Information Security
Pruning Unsafe Tickets: A Resource-Efficient Framework for Safer and More Robust LLMs
Apr 17, 2026Wai Man Si, Mingjie Li, Michael Backes +1LLM Pruning LLM Safety Alignment
CISPA Helmholtz Center for Information Security

May 7, 2026, cs.AIJ/K move · Enter open · S save

Chain of Risk: Safety Failures in Large Reasoning Models and Mitigation via Adaptive Multi-Principle Steering

Xiaomin Li, Jianheng Hou, Zheyuan Deng, Zhiwei Zhang+5

Harvard University · University of Southern California · Brown University+6

LLM Safety LLM Safety Evaluation Activation Steering Large Reasoning Models

Open paper AskPDF arXiv

Large reasoning models (LRMs) increasingly expose chain-of-thought-like reasoning for transparency, verification, and deliberate problem solving. This creates a safety blind spot: harmful or policy-violating content may appear in reasoning traces even when final answers appear safe. We test whether final-answer safety is a sufficient proxy for the full reasoning-answer trajectory by scoring both stages under a unified twenty-principle safety rubric. Using prompts from seven public harmfulness and jailbreak sources, plus four out-of-distribution (OOD) sources, we evaluate 15 open-weight and API-based LRMs across 41K prompts per model. Reasoning traces consistently reveal additional safety risks beyond final answers, especially in high-severity stage-wise failures: leak cases, where unsafe reasoning precedes a safe-looking answer, and escape cases, where benign-looking reasoning precedes an unsafe final response. Principle-level analysis shows that risk concentrates in misinformation, legal compliance, discrimination, physical harm, and psychological harm. We further propose adaptive multi-principle steering, a white-box test-time mitigation that learns one unsafe-to-safe activation direction per safety principle and activates only directions whose current hidden state is closer to the unsafe than safe centroid. On three steerable open reasoning models, adaptive steering reduces unsafe counts in both reasoning traces and final answers on held-out and OOD benchmarks. DeepSeek-R1-Qwen-7B achieves a 40.8% average unsafe-count reduction while retaining 97.7% macro-averaged accuracy on BBH, GSM8K, and MMLU. These results suggest that LRM safety should be evaluated and mitigated over the full exposed reasoning-answer trajectory, not only at the final-answer stage.

Momentum

Related topics

Latest papers 272