Image de couverture de livre

Detecting Safety Violations Across Many Agent Traces

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Stein, Adam, Brown, Davis, Hassani, Hamed, Naik, Mayur, Wong, Eric
Format:	Preprint
Publié:	2026
Sujets:	Artificial Intelligence Computation and Language
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Once Upon an Input: Reasoning via Per-Instance Program Synthesis
par: Stein, Adam, et autres
Publié: (2025)

Adaptively profiling models with task elicitation
par: Brown, Davis, et autres
Publié: (2025)

Do We Need Frontier Models to Verify Mathematical Proofs?
par: Naik, Aaditya, et autres
Publié: (2026)

Towards Compositionality in Concept Learning
par: Stein, Adam, et autres
Publié: (2024)

Evaluating the Performance of Large Language Models via Debates
par: Moniri, Behrad, et autres
Publié: (2024)

Instruction Following by Principled Boosting Attention of Large Language Models
par: Guardieiro, Vitoria, et autres
Publié: (2025)

BrowserArena: Evaluating LLM Agents on Real-World Web Navigation Tasks
par: Anupam, Sagnik, et autres
Publié: (2025)

ReasoningShield: Safety Detection over Reasoning Traces of Large Reasoning Models
par: Li, Changyi, et autres
Publié: (2025)

One-Shot Safety Alignment for Large Language Models via Optimal Dualization
par: Huang, Xinmeng, et autres
Publié: (2024)

Many-Tier Instruction Hierarchy in LLM Agents
par: Zhang, Jingyu, et autres
Publié: (2026)

Delta Activations: A Representation for Finetuned Large Language Models
par: Xu, Zhiqiu, et autres
Publié: (2025)

Lobster: A GPU-Accelerated Framework for Neurosymbolic Programming
par: Biberstein, Paul, et autres
Publié: (2025)

Governed Memory: A Production Architecture for Multi-Agent Workflows
par: Taheri, Hamed
Publié: (2026)

Multi-Agent LLMs for Generating Research Limitations
par: Azher, Ibrahim Al, et autres
Publié: (2025)

TraceSIR: A Multi-Agent Framework for Structured Analysis and Reporting of Agentic Execution Traces
par: Yang, Shu-Xun, et autres
Publié: (2026)

In Machina N400: Pinpointing Where a Causal Language Model Detects Semantic Violations
par: Zacharopoulos, Christos-Nikolaos, et autres
Publié: (2025)

An Empirical Study of Many-to-Many Summarization with Large Language Models
par: Wang, Jiaan, et autres
Publié: (2025)

Leveraging LLMs for Synthesizing Training Data Across Many Languages in Multilingual Dense Retrieval
par: Thakur, Nandan, et autres
Publié: (2023)

Agent-Based Detection and Resolution of Incompleteness and Ambiguity in Interactions with Large Language Models
par: Naik, Riya, et autres
Publié: (2025)

Policy-Invisible Violations in LLM-Based Agents
par: Wu, Jie, et autres
Publié: (2026)

Many-Turn Jailbreaking
par: Yang, Xianjun, et autres
Publié: (2025)

Personality Expression Across Contexts: Linguistic and Behavioral Variation in LLM Agents
par: Han, Bin, et autres
Publié: (2026)

It's Not the Capability: Harness Sensitivity Is Non-Monotone Across LLM Agent Tiers
par: Cho, Yong-eun
Publié: (2026)

Multimodal Situational Safety
par: Zhou, Kaiwen, et autres
Publié: (2024)

Why Do Safety Guardrails Degrade Across Languages?
par: Zhang, Max, et autres
Publié: (2026)

An AI-Based Behavioral Health Safety Filter and Dataset for Identifying Mental Health Crises in Text-Based Conversations
par: Nelson, Benjamin W., et autres
Publié: (2025)

One Patient, Many Contexts: Scaling Medical AI with Contextual Intelligence
par: Li, Michelle M., et autres
Publié: (2025)

Cultural Compass: A Framework for Organizing Societal Norms to Detect Violations in Human-AI Conversations
par: Cheng, Myra, et autres
Publié: (2026)

ToolSword: Unveiling Safety Issues of Large Language Models in Tool Learning Across Three Stages
par: Ye, Junjie, et autres
Publié: (2024)

Measuring Competency, Not Performance: Item-Aware Evaluation Across Medical Benchmarks
par: Luo, Zhimeng, et autres
Publié: (2025)

CRASH: Cognitive Reasoning Agent for Safety Hazards in Autonomous Driving
par: Silva, Erick, et autres
Publié: (2026)

R-Judge: Benchmarking Safety Risk Awareness for LLM Agents
par: Yuan, Tongxin, et autres
Publié: (2024)

The Fragility of Chain-of-Thought Monitoring Across Typologically Diverse Languages
par: Onyame, Eric, et autres
Publié: (2026)

Memory Transfer Learning: How Memories are Transferred Across Domains in Coding Agents
par: Kim, Kangsan, et autres
Publié: (2026)

When Only the Final Text Survives: Implicit Execution Tracing for Multi-Agent Attribution
par: Nian, Yi, et autres
Publié: (2026)

CatRAG: Functor-Guided Structural Debiasing with Retrieval Augmentation for Fair LLMs
par: Ranjan, Ravi, et autres
Publié: (2026)

LettuceDetect: A Hallucination Detection Framework for RAG Applications
par: Kovács, Ádám, et autres
Publié: (2025)

Why Do LLM-based Web Agents Fail? A Hierarchical Planning Perspective
par: Aghzal, Mohamed, et autres
Publié: (2026)

Alignment Backfire: Language-Dependent Reversal of Safety Interventions Across 16 Languages in LLM Multi-Agent Systems
par: Fukui, Hiroki
Publié: (2026)

CRMArena-Pro: Holistic Assessment of LLM Agents Across Diverse Business Scenarios and Interactions
par: Huang, Kung-Hsiang, et autres
Publié: (2025)