Unsafe

Follow this topic Ask about this topic

Momentum

2 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

Jul 6Week of Sep 21

Latest papers 28

All topics

Hard-Gate Candidacy in a Deployed Validator Suite
Sep 30, 2026Xin XuGating Validation
Carnegie Mellon University
Inoculation Midtraining with Learned Neologisms
Sep 14, 2026Kyle O'Brien, Edward James Young, Puria Radmard +4Large Language Model Safety Inoculation Prompting
Geodesic Research · OpenAI · UK AI Security Institute
ActSafeGuard: Differentiable and Training-Aligned Constraint Enforcement for Flow-Matching Policies
Sep 10, 2026Jianming Ma, Rongjun Jin, Xiaxi Si +3Safety Constraints Safer
1Shanghai Jiao Tong University · 2Shanghai Innovation Institute
Influence Is Not Authority: When Causal Guardrail Signals Make Legitimate Tool Use Look Like an Attack in Tool-Using LLM Agents
Aug 30, 2026Tanzim Ahad, Ismail Hossain, Md Jahangir Alam +3Controlling Tool Use Guardrail
Department of Computer Science, University of Texas at El Paso, TX, USA 79902 · Department of Computer Science and Engineering, New Mexico Tech, Socorro, NM, USA 87801 · University of Illinois Urbana-Champaign, IL, USA
DreamGuard: Efficient Runtime Guardrail for LLM Agents via Risk-Aware World Model
Aug 6, 2026Wenhao Lin, Chenyu Yu, Xingwei Lin +6Runtime Safety Filtering Large Language Model Agents
Zhejiang University · Nanjing University of Posts and Telecommunications · Sun Yat-sen University
Mind the Gap: Zero-Query Jailbreaks via Filter-Generator Discrepancy in Text-to-Image Systems
Aug 2, 2026Wanguang Li, Zhaoxin Wang, Handing WangAdversarial Prompts Jailbreak Success Rates
School of Artificial Intelligence, Xidian University, Shaanxi, China
Single Canonical Prompts Underestimate LLM Safety's Surface-Form Sensitivity
Aug 1, 2026Yongxi Zhou, Junwei Yao, Yuanzhe Liu +4Large Language Model Safety Unsafe
1Northeastern University, Massachusetts, USA · 2Georgia Institute of Technology, Georgia, USA · University of Southern California, California, USA
Falling Behind Drives Unsafe Development in an Idealised AI Race Experiment
Jul 28, 2026Elias Fernández Domingos, The Anh HanArtificial Intelligence Safety Artificial Intelligence Systems
AI-Lab, Vrije Universiteit Brussel, Pleinlaan 9, Brussels, 1050, Belgium. · MLG, Université Libre de Bruxelles, Boulevard du Triomphe, Brussels, 1050, Belgium. · School of Computing, Engineering and Digital Technologies, Teesside University, Teesside University, Middlesbrough, United Kingdom. +1
Adjustment Speed as a Safety Constraint for Nonstationary Reinforcement Learning
Jul 21, 2026Timothy TomashevskiySafety Constraints Offline Reinforcement Learning
Department of Computing and Software McMaster University Hamilton, Ontario, Canada
Unsafe at any AUC: Unlearned Lessons from Sociotechnical Disasters for Responsible AI
Jul 15, 2026Joshua A. Kroll, Andrew Smart, R. Stuart Geiger +1Socio-Technical Systems Artificial Intelligence Risk
Naval Postgraduate School · Google Research · University of California, San Diego +1
Avoiding unsafe sets when training with Langevin Dynamics
Jul 8, 2026Adam M. ObermanAnisotropic Loss Landscapes Langevin Dynamics
Department of Mathematics and Statistics, McGill University, Montreal, QC, Canada; Mila, Quebec AI Institute; and LawZero
Safe Alone, Unsafe Together: Safeguarding Against Implicit Toxicity When Benign Images Combine
Jul 1, 2026Jiaxian Lv, Shiyao Cui, Yingkang Wang +3Content Moderation Toxicity
The Conversational AI (CoAI) Group, DCST, Tsinghua University
Safe-RULE: Safe Reinforcement UnLEarning
Jun 8, 2026Shixiong Jiang, Taozheng Zhu, Fanxin KongModel-Based Reinforcement Learning Unsafe
Department of Computer Science and Engineering, University of Notre Dame
AICompanionBench: Benchmarking LLMs-as-Judges for AI Companion Safety
Jun 3, 2026Yanjing Ren, Reza Ebrahimi, TengTeng MaArtificial Intelligence Companions Llm-As-A-Judge
The School of Information Systems University of South Florida Tampa, US
Propagating Unsafe Actions in LLM Controlled Multi-Robot Collaboration via Single Robot Compromise
May 15, 2026Zhen Huang, Zhihuang Liu, Mengxuan Luo +2Multi-Robot Systems Safety Alignment
College of Computer Science and Technology, National University of Defense Technology
3DEditSafe: Defending 3D Editing Pipelines from Unsafe Generation
May 14, 2026Nicole Meng, Zheyuan Liu, Meng Jiang +13D Editing 3D Generation
Tufts University Medford, MA, 02155 · University of Notre Dame Notre Dame, IN, 46556
Safe, or Simply Incapable? Rethinking Safety Evaluation for Phone-Use Agents
May 8, 2026Zhengyang Tang, Yi Zhang, Chenxin Li +18Safety Benchmarks Unsafe
1Tencent Hunyuan · 2The Chinese University of Hong Kong, Shenzhen · 3Tsinghua University
Good in Bad (GiB): Sifting Through End-user Demonstrations for Learning a Better Policy
May 2, 2026Noushad Sojib, Ola Ghattas, Momotaz BegumImitation Learning Bad
Department of Computer Science, University of New Hampshire, USA
Ambient Persuasion in a Deployed AI Agent: Unauthorized Escalation Following Routine Non-Adversarial Content Exposure
Apr 29, 2026Diego F. Cuadros, Abdoul-Aziz MaigaAgentic Deployments Escalation
Digital Epidemiology Laboratory, Digital Futures, University of Cincinnati, Cincinnati, OH, USA · Center for Humanities and Technology (CHaT), University of Cincinnati, Cincinnati, OH, USA · Norwegian University of Science and Technology, Trondheim, Norway
The Safety-Aware Denoiser for Text Diffusion Models
Apr 28, 2026Amman Yusuf, Zhejun Jiang, Mijung ParkDiffusion Language Models Large Language Model Safety
Department of Computer Science, The University of British Columbia, Vancouver, Canada.
From Concept-Aligned Tokens to Vulnerable Features: Mechanistic Localization of Jailbreaks
Apr 25, 2026Nilanjana Das, Mathew Dawit, Aman Chadha +1Jailbreak Success Rates Unsafe
UMBC, Maryland, U.S.A. · Apple, California, U.S.A.
RedVLA: Physical Red Teaming for Vision-Language-Action Models
Apr 24, 2026Yuhao Zhang, Borong Zhang, Jiaming Fan +4Red-Teaming Unsafe
Institute for AI, Peking University · State Key Laboratory of General Artificial Intelligence, Peking University
Pruning Unsafe Tickets: A Resource-Efficient Framework for Safer and More Robust LLMs
Apr 17, 2026Wai Man Si, Mingjie Li, Michael Backes +1Large Language Model Safety Unsafe
CISPA Helmholtz Center for Information Security
Subliminal Transfer of Unsafe Behaviors in AI Agent Distillation
Apr 16, 2026Jacob Dang, Brian Y. Xie, Omar G. YounisUnsafe Dataset Distillation
UCLA · Santa Monica College · Mila, Silverstream AI
Unsafe by Reciprocity: How Generation-Understanding Coupling Undermines Safety in Unified Multimodal Models
Mar 28, 2026Kaishen Wang, Heng HuangUnsafe Large Language Model Uncertainty
University of Maryland, College Park, MD 20742, USA
Measuring Harmfulness of Computer-Using Agents
Jul 31, 2025Aaron Xuxiang Tian, Ruofan Zhang, Janet Tang +3Large Language Model Safety Unsafe
Independent Researcher · Arizona State University · Gradient +1
Learning-based Adaptive Safety-Critical Control With Evolving Unsafe Regions
Mar 19, 2025Songqiao Hu, Zidong Wang, Zeyi Liu +2Control Barrier Functions Safety-Critical Scenarios
Department of Automation and the Institute for Embodied Intelligence and Robotics, Tsinghua University, Beijing 100084, China · Department of Computer Science, Brunel University London, UB8 3PH Uxbridge, U.K. · State Key Laboratory of Multimodal Artificial Intelligence Systems, Beijing Engineering Research Center of Intelligent Systems and Technology, Institute of Automation, Chinese Academy of Sciences, Beijing, China
SaFeR-Steer: Evolving Multi-Turn MLLMs via Synthetic Bootstrapping and Feedback Dynamics
Date pendingHaolong Hu, Hanyu Li, Tiancheng He +6Mllm-Based Large Language Model Alignment
Huazhong University of Science and Technology · Beijing University of Posts and Telecommunications · West China Biomedical Big Data Center, Sichuan University +2

Sep 30, 2026, cs.LGJ/K move · Enter open · S save

Hard-Gate Candidacy in a Deployed Validator Suite

Xin Xu

Carnegie Mellon University

Gating Validation Runtime Enforcement Candidate Unsafe

Open paper AskPDF arXiv

Before a validator can be promoted to a hard gate on a deployment pipeline, it has to be shown that its firing separates outputs that reach users in working order from those that do not. We run that screen on 13 validators in a deployed generative agent, against 550 runtime and 350 static builds labelled by downstream outcome, and report each check's marginal separation

J=\mathrm{TPR}-\mathrm{FPR}

with Newcombe intervals and Fisher exact tests. Two checks survive correction for multiple comparisons, two more are nominal only, and the remaining nine are not distinguishable from zero, three of them because they never fired on any sampled build. Execution itself is not random with respect to the property being gated, and this replicates: across four runs covering 1,867 builds and ten distinct runtime checks, probes were skipped on 144 of 895 broken builds and 1 of 972 acceptable builds (per-run rates 15.6% to 16.6% against at most 0.3%), every skip carrying the same unsafe-to-probe reason. Because a skipped check is recorded as a pass, this imposes a ceiling that no check quality can lift: a check that needs a live artifact cannot operationally detect more than about 84% of broken builds in this harness. For the one check with construct-specific labels, a detector built for blank output fires on 0 of 90 human-labelled blank builds (95% upper bound on sensitivity 3.3%), and the global frame statistic it approximates separates the classes only weakly (AUC 0.59), so the gap is not a threshold that needs tuning. The same gap appears one layer up: on a census of tens of thousands of judge-scored builds, 32.5% of rejections carry no recorded issue at all. We argue that evaluation records must distinguish a check that ran and passed from one that did not run, must carry the evidence for a rejection, and that an inventory of checks is not evidence about a gate.

Momentum

Related topics

Latest papers 28