Copertina

When Actions Go Off-Task: Detecting and Correcting Misaligned Actions in Computer-Use Agents

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Ning, Yuting, Jones, Jaylen, Zhang, Zhehao, Ye, Chentao, Ruan, Weitong, Li, Junyi, Gupta, Rahul, Sun, Huan
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

When Benign Inputs Lead to Severe Harms: Eliciting Unsafe Unintended Behaviors of Computer-Use Agents
di: Jones, Jaylen, et al.
Pubblicazione: (2026)

RedTeamCUA: Realistic Adversarial Testing of Computer-Use Agents in Hybrid Web-OS Environments
di: Liao, Zeyi, et al.
Pubblicazione: (2025)

Preemptive Detection and Correction of Misaligned Actions in LLM Agents
di: Fang, Haishuo, et al.
Pubblicazione: (2024)

AmpleGCG-Plus: A Strong Generative Model of Adversarial Suffixes to Jailbreak LLMs with Higher Success Rates in Fewer Attempts
di: Kumar, Vishal, et al.
Pubblicazione: (2024)

A Multi-Aspect Framework for Counter Narrative Evaluation using Large Language Models
di: Jones, Jaylen, et al.
Pubblicazione: (2024)

When Actions Teach You to Think: Reasoning-Action Synergy via Reinforcement Learning in Conversational Agents
di: Rawat, Mrinal, et al.
Pubblicazione: (2025)

When Stopping Requires Going: Physiological Similarities Between Action Cancellation and the Cancellation of Action Cancellation
di: Simon Weber, et al.
Pubblicazione: (2025)

Gradients as an Action: Towards Communication-Efficient Federated Recommender Systems via Adaptive Action Sharing
di: Lu, Zhufeng, et al.
Pubblicazione: (2025)

IntentScore: Intent-Conditioned Action Evaluation for Computer-Use Agents
di: Chen, Rongqian, et al.
Pubblicazione: (2026)

ROGUE: Misaligned Agent Behavior Arising from Ordinary Computer Use
di: Tien, Jeremy, et al.
Pubblicazione: (2026)

Action Contextualization: Adaptive Task Planning and Action Tuning using Large Language Models
di: Gupta, Sthithpragya, et al.
Pubblicazione: (2024)

UltraCUA: A Foundation Model for Computer Use Agents with Hybrid Action
di: Yang, Yuhao, et al.
Pubblicazione: (2025)

How Catastrophic is Your LLM? Certifying Risk in Conversation
di: Wang, Chengxiao, et al.
Pubblicazione: (2025)

Prioritize Team Actions: Multi-Agent Temporal Logic Task Planning with Ordering Constraints
di: Ye, Bowen, et al.
Pubblicazione: (2024)

Efficient Multi-Task Reinforcement Learning via Task-Specific Action Correction
di: Feng, Jinyuan, et al.
Pubblicazione: (2024)

World-Model-Augmented Web Agents with Action Correction
di: Shen, Zhouzhou, et al.
Pubblicazione: (2026)

Action-Graph Policies: Learning Action Co-dependencies in Multi-Agent Reinforcement Learning
di: Gupta, Nikunj, et al.
Pubblicazione: (2026)

CAPTCHA Solving for Native GUI Agents: Automated Reasoning-Action Data Generation and Self-Corrective Training
di: Chen, Yuxi, et al.
Pubblicazione: (2026)

Return-to-Go Is More Than a Number: Q-Guided Alignment for Return-Conditioned Supervised Learning
di: Yang, Yuxiao, et al.
Pubblicazione: (2026)

Off-Shell Strings I: S-matrix and Action
di: Ahmadain, Amr, et al.
Pubblicazione: (2022)

Learning Action Embeddings for Off-Policy Evaluation
di: Cief, Matej, et al.
Pubblicazione: (2023)

Q-LINK: Quantum Layerwise Information Residual Network via a Messenger Qubit for Barren Plateaus Mitigation
di: Yi, Zhehao, et al.
Pubblicazione: (2026)

The PID Controller Strikes Back: Classical Controller Helps Mitigate Barren Plateaus in Noisy Variational Quantum Circuits
di: Yi, Zhehao, et al.
Pubblicazione: (2025)

Geometric Optimization on Lie Groups: A Lie-Theoretic Explanation of Barren Plateau Mitigation for Variational Quantum Algorithms
di: Yi, Zhehao, et al.
Pubblicazione: (2025)

Neural-network Generated Quantum State Can Mitigate the Barren Plateau in Variational Quantum Circuits
di: Yi, Zhehao, et al.
Pubblicazione: (2024)

The Basis and Applications of the Action Fluency and Action Naming Tasks
di: Bárbara Costa Beber
Pubblicazione: (2014)

EcoAct: Economic Agent Determines When to Register What Action
di: Zhang, Shaokun, et al.
Pubblicazione: (2024)

Customize Multi-modal RAI Guardrails with Precedent-based predictions
di: Yang, Cheng-Fu, et al.
Pubblicazione: (2025)

When to Trust Imagination: Adaptive Action Execution for World Action Models
di: Wang, Rui, et al.
Pubblicazione: (2026)

GUI Action Narrator: Where and When Did That Action Take Place?
di: Wu, Qinchen, et al.
Pubblicazione: (2024)

Video-Based Reward Modeling for Computer-Use Agents
di: Song, Linxin, et al.
Pubblicazione: (2026)

When Parallelism Pays Off: Cohesion-Aware Task Partitioning for Multi-Agent Coding
di: Yang, Xu, et al.
Pubblicazione: (2026)

CAMMARL: Conformal Action Modeling in Multi Agent Reinforcement Learning
di: Gupta, Nikunj, et al.
Pubblicazione: (2023)

ALOE: Action-Level Off-Policy Evaluation for Vision-Language-Action Model Post-Training
di: Yang, Rushuai, et al.
Pubblicazione: (2026)

QUEST: Training Frontier Deep Research Agents with Fully Synthetic Tasks
di: Xie, Jian, et al.
Pubblicazione: (2026)

Adapting Short-Term Transformers for Action Detection in Untrimmed Videos
di: Yang, Min, et al.
Pubblicazione: (2023)

Action Recommendations for Sequentially Rational Strategic Agents
di: Sun, Renyan, et al.
Pubblicazione: (2026)

Action Controlled Paraphrasing
di: Shi, Ning, et al.
Pubblicazione: (2024)

When Actions Disappear: Adversarial Action Removal in Self-Play Reinforcement Learning
di: Kujur, Arahan
Pubblicazione: (2026)

NEBULA: Do We Evaluate Vision-Language-Action Agents Correctly?
di: Peng, Jierui, et al.
Pubblicazione: (2025)