LLM Agent Safety

LLM: Large Language Model

Follow this topic Ask about this topic

Momentum

24 papers in the last four weeks, up 50% on the four weeks before. 0.2% of all new papers.

Jul 13Week of Sep 28

Latest papers 193

All topics

What to Admit and How to Present: Governing Persistent Memory in LLM Agents
Oct 8, 2026Chang Liu, Deliang DingAgent Memory Management Persistent Memory for Language Models
Pazhou Laboratory Guangzhou, China
POLAR: Ontology-Guided Risk Prevention for Tool-Calling LLM Agents
Oct 6, 2026Yunju Kang, Seonghyeon Cho, Irene Li +2LLM Guardrails Runtime Enforcement for AI Agents
Soongsil University · Tokyo University · Jeju National University
DecepEval: A Benchmark for Evaluating Deception in LLM Agents
Oct 6, 2026Yiming Xu, Hongyue Yu, Beihua Yang +8Deception in Language Models LLM Agent Evaluation
Xi’an Jiaotong University · University of Virginia · Griffith University +2
BazaarBench: Delegation Safety in Decentralized C2C Marketplaces Run by LLM Agents
Oct 5, 2026Ziyan Wang, Shuqing Shi, James Oldfield +5AI Agent Safety Benchmarks LLM Agent Safety
King’s College London · Institute for Decentralized AI · University of Oxford +1
G-CARB: Graph-Localized Conformal Agent Risk Budget for Compositional Harm
Oct 4, 2026Zijun Yu, Yu Gu, Vahid Partovi Nia +1AI Agent Safety Conformal Risk Control
Department of Mathematics and Statistics McGill University Montréal, QC, Canada · Department of Mathematics and Industrial Engineering Polytechnique de Montréal Montréal, QC, Canada
Representation Transitions Reveal Emerging Safety Risks in Multi-Turn LLM Agents
Sep 30, 2026Haoyu Wang, Wei Zhao, Yedi Zhang +2AI Agent Monitoring LLM Agent Safety
Singapore Management University · National University of Singapore
ActionGuard: Tool Call Authorization under Poisoned Skills
Sep 30, 2026Jihun Han, Yejin Jang, Byung Il Kwak +1Runtime Enforcement for AI Agents Prompt Injection Defense
Korea University Republic of Korea
Janus: Evidence-Before-Effect Sagas and Offline-Verifiable Provenance for Agentic LLMs
Sep 29, 2026Mustafa ArslanLLM Auditing AI Agent Governance
Independent Researcher Istanbul, Türkiye
SafeCoEvo: Co-Evolving Safety Harnesses and Guards for LLM Agents at Test-Time
Sep 29, 2026Yu Cheng, Yongkang Hu, Shuaijie Ma +12Continual Test-Time Adaptation LLM Agents
East China Normal University · Shanghai Innovation Institute · Xiamen University +4
SCOUT: Synergizing Reasoning and Tool-Use for Computer-Use Safety
Sep 28, 2026Jianxing Chen, Xiao Yu, Shipra Agrawal +1AI Agent Safety AI Agent Safety Benchmarks
Columbia University
Tool Mediation Alters Refusal Mechanisms in Large Language Models
Sep 28, 2026Abel Rodríguez, Giuseppe Garofalo, Lieven Desmet +1Language Model Safety Evaluation LLM Agent Safety
DistriNet, KU Leuven
Jailbreak Context Lingers: Divergent Safety Routing and Its Cross-Task Predictability in Tool Agents
Sep 28, 2026Xi Wang, Songlei Jian, Yiming Zhang +5LLM Safety Evaluation Causal Interventions in Language Models
National University of Defense Technology
PROACT-Agent: Progressive Runtime Oversight and Active Circuit-breaking for Real-Time Safety
Sep 28, 2026Ding Jia, Wei Liu, Xianglong Du +5LLM Guardrails Runtime Enforcement for AI Agents
State Key Laboratory of Intelligent Vehicle Safety Technology, Changan Automobile · Independent Researcher
Population Physics, Population Problems: Safety and Emergence in LLM Societies
Sep 27, 2026Adrian de WynterMulti-Agent LLM Systems Large Language Model-Based Social Simulation
Microsoft & The University of York
AgentBoundary: Counterfactual Evaluation of Safety in Tool-Using LLM Agents
Sep 27, 2026Tianzhuo Yang, Zirui Mi, Yantao Huang +4LLM Agent Evaluation LLM Agents
Peking University · Beijing Academy of Artificial Intelligence
Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure
Sep 24, 2026David Schmotz, Derck Prinzhorn, Luca Beurer-Kellner +3Runtime Enforcement for AI Agents AI Agent Safety Benchmarks
ELLIS Institute Tübingen · Max Planck Institute for Intelligent Systems · Tübingen AI Center +3
SkinAgent AI: A Safety-Grounded Multimodal Agentic Framework for Non-Diagnostic Skincare Support
Sep 24, 2026Muhammad Muhtasim Shahriar, Abdullah Mohammad Sayem, Tze Hui Liew +2LLM Agent Orchestration LLM Agent Evaluation
Department of Computer Science, International Islamic University Chittagong, Chittagong, Bangladesh · Centre for Intelligent Cloud Computing, Centre of Excellence for Advanced Cloud, Faculty of Information Science and Technology, Multimedia University, Melaka, Malaysia · Multimedia University, Jalan Ayer Keroh Lama, 75450 Bukit Beruang, Melaka, Malaysia +1
PASTABench: Proactive Assessment of Sequential Trajectories for Agent Safety
Sep 23, 2026Jiapeng Sun, Yujin Zhou, Han Zhu +4AI Agent Safety Benchmarks Inference-Time Intervention
The Hong Kong University of Science and Technology, Hong Kong, China · Peking University, Beijing, China
Emergent Collusion in Long-Horizon LLM Agent Interaction
Sep 21, 2026Xinrui Shi, Yanzhe Zhang, Diyi YangMulti-Agent Coordination LLM Agent Safety
Stanford University · Georgia Tech
Indirect tipping: a social attack surface in AI agent populations
Sep 21, 2026Ariel Flint, Luca Maria Aiello, Sara M. Constantino +2Multi-Agent Coordination AI Agent Security
Department of Mathematics, City St George’s, University of London, UK. · IT University of Copenhagen, Denmark. · Pioneer Centre for AI, Copenhagen, Denmark. +2
MAGS: Multi-agent Auto-formalization Guarantees Safety for Agentic Outputs
Sep 16, 2026Albert Wu, Nicholas Roberts, Tzu-Heng Huang +5Multi-Agent LLM Systems Formal Verification
University of Wisconsin-Madison · Princeton University
Symbolic Temporal Supervision of LLM Agents Using Contracts
Sep 16, 2026Yifeng Xiao, Pierluigi NuzzoRuntime Enforcement for AI Agents LLM Agent Safety
Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, CA, USA
BLINDSPOT: A Benchmark for Safety and Refusal Calibration in Long-Horizon Tool-Using Agents
Sep 14, 2026Sadia Asif, Mohammad Mohammadi Amiri, Momin Abbas +2Long-Horizon Agent Evaluation AI Agent Safety Benchmarks
Rensselaer Polytechnic Institute · IBM Research
Why LLM Agents Collapse Without Oversight: The Enforcement Gap as the Mechanism Behind Emergence World Failures
Sep 14, 2026Yuhang WangLLM Auditing LLM Agent Security
Fudan University
ActGuard: Pre-execution Action Auditing against Indirect Prompt Injection in LLM Agents
Sep 14, 2026Bingzheng Wang, Xiaoyan Gu, Wentao Wang +3Runtime Enforcement for AI Agents Prompt Injection Defense
Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China · School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China · Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China 2School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China 3Key Laboratory of Cyberspace Security Defense, Beijing, China +3
Revoked but Still Authoritative: An Empirical Study of Revocation Enforcement in Agent-Memory Systems
Sep 8, 2026Yi Ting Shen, Kentaroh Toyoda, Alex LeungLLM Agent Evaluation Runtime Enforcement for AI Agents
Vulcan Research, AIFT Singapore
SchemeArena: Factorized Stress Testing of Scheming in LLM Agents
Sep 8, 2026Jie Ruan, Inderjeet Nair, Amy Liu +3LLM Agent Evaluation AI Agent Monitoring
Computer Science and Engineering University of Michigan Ann Arbor, MI 48109, USA
How Long Until Your Robot Ignores You? A Safety Benchmark for LLM Orchestrators in Human-Humanoid Collaboration
Sep 7, 2026Aulon Bajrami, Mohamed Elshamouty, Werner KrausRobot Safety Human-Robot Collaboration
Fraunhofer IPA Stuttgart, Germany
HarvestBench: Measuring Whether LLM Agents Will Pay to Avoid Killing Animals
Sep 3, 2026Jasmine Brazilek, Miles Tidmarsh, Matthias Endres +2Moral Reasoning in Language Models LLM Safety Alignment
Compassion Aligned Machine Learning (CaML) · Department of Economics, University of Warwick
SafeEvolve: Harness-Policy Co-Evolution from Agent Experience for Safety Alignment
Sep 2, 2026Qinghua Mao, Wanying Qu, Dadi Guo +8LLM Safety Alignment LLM Agent Safety
Shanghai AI Laboratory · SJTU · Fudan University +2

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

What to Admit and How to Present: Governing Persistent Memory in LLM Agents

Chang Liu, Deliang Ding

Pazhou Laboratory Guangzhou, China

Agent Memory Management Persistent Memory for Language Models LLM Agent Safety Privacy Leakage in Language Models

Open paper AskPDF arXiv

Persistent memory can improve personalization in LLM agents but can also induce sycophancy and cross-domain leakage. We distinguish two governance decisions: admission, which determines what recalled information enters the working context, and presentation, which determines how admitted information is expressed. We implement two inference-time designs without retraining: factor-compiled admission (FC), which assesses whole memory entries, and permission-semantic admission (PS), which decomposes entries into typed units; both translate adjudicated attributes into eligibility decisions via deterministic policies. We evaluate on a four-backbone development suite and an external benchmark with four tasks of 300 samples each. Relative to verbatim injection, FC and PS reduce pooled judge-assessed failure rates on the external benchmark by 6.7 and 8.8 percentage points (p = 2.7e-7 and 4.1e-12), and development-set cross-domain leakage falls by up to 29.5 percentage points. A query-conditioned gating baseline shows no significant change in objective-fact failure or pooled failure. Under matched admission budgets, PS outperforms random and relevance-based selection on external objective-fact judgment after Holm correction. Holding presentation fixed, tightening admission cuts cross-domain failure by a further 17.5 percentage points (p = 1.6e-4); in contrast, no comparison between two renderings of identical adjudicated outputs survives multiple-comparison correction. Both designs increase personalization failures, and PS misses the preregistered improvement and personalization-preservation criteria. These results support evaluating admission and presentation separately: selection quality provides task-specific safety gains, while preserving beneficial memory use remains unresolved.

Momentum

Related topics

Latest papers 193