LLM Safety Alignment

LLM: Large Language Model

Follow this topic Ask about this topic

Momentum

25 papers in the last four weeks, up 14% on the four weeks before. 0.2% of all new papers.

Jul 13Week of Sep 28

Latest papers 236

All topics

Latent Personality Alignment: Improving Harmlessness Without Mentioning Harms
May 8, 2026Linh Le, David Williams-King, Mohamed Amine Merzouk +2Adversarial Training Neural Network Robustness
McGill University · Mila – Quebec AI Institute · ERA +1
The Attacker in the Mirror: Breaking Self-Consistency in Safety via Anchored Bipolicy Self-Play
May 8, 2026Gabriele La Malfa, Emanuele La Malfa, Saar Cohen +4Self-Play RL Language Model Safety Evaluation
Department of Informatics, King’s College London · Department of Computer Science, University of Oxford · Institute for Decentralized AI +1
Beyond "I cannot fulfill this request": Alleviating Rigid Rejection in LLMs via Label Enhancement
May 8, 2026Ying Zhang, Congyu Qiao, Xin Geng +1LLM Safety Alignment LLM Refusal Behavior
Hard to Read, Easy to Jailbreak: How Visual Degradation Bypasses MLLM Safety Alignment
May 8, 2026Zhixue Song, Boyan Han, Yiwei Wang +1VLM Robustness Adversarial Attacks on VLMs
1AGI Lab, Westlake University, China · University of California, Merced, USA
Measuring Evaluation-Context Divergence in Open-Weight LLMs: A Paired-Prompt Protocol with Pilot Evidence of Alignment-Pipeline-Specific Heterogeneity
May 7, 2026Florian A. D. Burnat, Brittany I. DavidsonLLM Safety Benchmarks LLM Evaluation
University of Bath
Safety Anchor: Defending Harmful Fine-tuning via Geometric Bottlenecks
May 7, 2026Guoxin Lu, Letian Sha, Qing Wang +4LLM Fine-Tuning LLM Safety Alignment
Nanjing University of Posts and Telecommunications, Nanjing, China.
Self-Mined Hardness for Safety Fine-Tuning
May 4, 2026Prakhar Gupta, Garv Shah, Donghua ZhangHard Negative Mining LLM Fine-Tuning
University of Michigan
Model Spec Midtraining: Improving How Alignment Training Generalizes
May 3, 2026Chloe Li, Nevan Wichers, Sara Price +2LLM Alignment LLM Safety Alignment
Anthropic Fellows Program · Anthropic
RefusalGuard: Geometry-Preserving Fine-Tuning for Safety in LLMs
May 3, 2026Sadia Asif, Mohammad Mohammadi AmiriFine-Tuning LLM Fine-Tuning
Department of Computer Science Rensselaer Polytechnic Institute Troy, NY, USA
Disentangling Intent from Role: Adversarial Self-Play for Persona-Invariant Safety Alignment
May 3, 2026Jiajia Li, Xiaoyu Wen, Zhongtian Ma +3LLM Alignment LLM Safety Alignment
Northwestern Polytechnical University · Shanghai AI Laboratory · Shanghai Jiao Tong University +1
Multilingual Safety Alignment via Self-Distillation
May 3, 2026Ruiyang Qin, Qingzhuo Wang, Dongrui Liu +3Multilingual Language Models LLM Safety Alignment
Tongji University · Shanghai AI Laboratory
A Theoretical Game of Attacks via Compositional Skills
May 1, 2026Xinbo Wu, Huan Zhang, Abhishek Umrawal +1Adversarial Prompt Generation Adversarial Attacks on LLMs
University of Illinois Urbana-Champaign, Urbana, IL, USA · Stony Brook University, Stony Brook, NY, USA.
Breaking Safety at the Token Boundary: How BPE Tokenization Creates Exploitable Gaps in LLM Alignment
May 1, 2026Tung-Ling Li, Hongliang Liu, Yuhao WuLanguage Model Safety Evaluation Adversarial Attacks on LLMs
Palo Alto Networks
ARMOR 2025: A Military-Aligned Benchmark for Evaluating Large Language Model Safety Beyond Civilian Contexts
Apr 30, 2026Sydney Johns, Heng Jin, Chaoyu Zhang +2LLM Safety Benchmarks LLM Safety Alignment
Virginia Polytechnic Institute and State University, Virginia, United States
Attention Is Where You Attack
Apr 30, 2026Aviral Srivastava, Sourav PandaSelf-Attention Adversarial Attacks
Amazon · The Pennsylvania State University
Tatemae: Detecting Alignment Faking via Tool Selection in LLMs
Apr 29, 2026Matteo Leonesi, Francesco Belardinelli, Flavio Corradini +1Deception in Language Models LLM Safety Alignment
Department of Computing Imperial College London
Test-Time Safety Alignment
Apr 28, 2026Baturay Saglam, Dionysis KalogeriasWord Embeddings LLM Safety Alignment
Department of Electrical and Computer Engineering Yale University
Frictive Policy Optimization for LLMs: Epistemic Intervention, Risk-Sensitive Control, and Reflective Alignment
Apr 28, 2026James Pustejovsky, Nikhil KrishnaswamyRL for Language Models Language Model Safety Evaluation
Brandeis University · Waltham, MA, USA · Colorado State University +1
Evaluating Risks in Weak-to-Strong Alignment: A Bias-Variance Perspective
Apr 28, 2026Hamid Osooli, Kareema Batool, Rick Gentry +3LLM Safety Alignment Weak-to-Strong Generalization
University of Illinois Urbana-Champaign · InstaDeep · Microsoft +1
Multilingual Refusal Alignment for Safer Large Language Models
Apr 24, 2026Aleksandra Krasnodębska, Wojciech Kusa, Aldo LipaniMultilingual Language Model Evaluation LLM Alignment
NASK National Research Institute, Warsaw, Poland · University College London, London, UK
Dialect vs Demographics: Quantifying LLM Bias from Implicit Linguistic Signals vs. Explicit User Profiles
Apr 22, 2026Irti Haq, Belén SaldíasLanguage Model Safety Evaluation Social Bias in Language Models
University of Washington, USA
SafeRedirect: Defeating Internal Safety Collapse via Task-Completion Redirection in Frontier LLMs
Apr 22, 2026Chao Pan, Yu Wu, Xin YaoLLM Guardrails LLM Safety Alignment
Department of Computer Science and Engineering, Southern University of Science and Technology, Shenzhen 518055, China · The Hong Kong Polytechnic University, Hong Kong, China · George Washington University +1
HarDBench: A Benchmark for Draft-Based Co-Authoring Jailbreak Attacks for Safe Human-LLM Collaborative Writing
Apr 21, 2026Euntae Kim, Soomin Han, Buru ChangLLM Safety Benchmarks Language Model Safety Evaluation
Korea University · Sogang University
Reasoning Structure Matters for Safety Alignment of Reasoning Models
Apr 21, 2026Yeonjun In, Wonjoong Kim, Sangwu Park +1LLM Safety Alignment Large Reasoning Models
KAIST
ARES: Adaptive Red-Teaming and End-to-End Repair of Policy-Reward System
Apr 20, 2026Jiacheng Liang, Yao Ma, Tharindu Kumarage +5Reward Modeling Adversarial Attacks on LLMs
Stony Brook University · Amazon Nova Responsible AI
Adversarial Arena: Crowdsourcing Data Generation through Interactive Competition
Apr 20, 2026Prasoon Goyal, Sattvik Sahai, Michael Johnston +14Synthetic Data Generation LLM Safety Alignment
Amazon Nova Responsible AI
SafeAnchor: Preventing Cumulative Safety Erosion in Continual Domain Adaptation of Large Language Models
Apr 20, 2026Dongxin Guo, Jikun Wu, Siu Ming YiuContinual Learning for LLMs LLM Fine-Tuning
The University of Hong Kong, Hong Kong, China · Brain Investing Limited, Hong Kong, China
Cat-DPO: Category-Adaptive Safety Alignment
Apr 19, 2026Tiankai Yang, Yi Nian, Xinyuan Li +3LLM Alignment Direct Preference Optimization
University of Southern California · 2Northwestern University
Continual Safety Alignment via Gradient-Based Sample Selection
Apr 19, 2026Thong Bach, Dung Nguyen, Thao Minh Le +1Continual Learning for LLMs LLM Fine-Tuning
Applied Artificial Intelligence Initiative (A2I2), Deakin University · Pennsylvania State University
Guardrails in Logit Space: Safety Token Regularization for LLM Alignment
Apr 19, 2026Thong Bach, Truyen TranLLM Fine-Tuning LLM Safety Alignment
Applied Artificial Intelligence Initiative (A2I2), Deakin University

May 8, 2026, cs.AIJ/K move · Enter open · S save

Latent Personality Alignment: Improving Harmlessness Without Mentioning Harms

Linh Le, David Williams-King, Mohamed Amine Merzouk, Aton Kamanda+1

McGill University · Mila – Quebec AI Institute · ERA+1

Adversarial Training Neural Network Robustness LLM Safety Alignment Adversarial Robustness of Language Models

Open paper AskPDF arXiv

Current adversarial robustness methods for large language models require extensive datasets of harmful prompts (thousands to hundreds of thousands of examples), yet remain vulnerable to novel attack vectors and distributional shifts. We propose Latent Personality Alignment (LPA), a sample-efficient defense that achieves robustness by training models on abstract personality traits rather than specific harmful behaviors. Using fewer than 100 trait statements and latent adversarial training, LPA achieves comparable attack success rates to methods trained on 150k+ examples, while maintaining superior utility. Critically, LPA generalizes better to unseen attack distributions, reducing misclassification rates by 2.6x compared to baseline across six harm benchmarks -- without ever seeing harmful examples during training. Our results demonstrate that personality-based alignment offers a principled approach to building robust defenses with minimal cost.

Momentum

Related topics

Latest papers 236