Imagen de Portada

Discovering Agentic Safety Specifications from 1-Bit Danger Signals

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autor principal:	Gallego, Víctor
Formato:	Preprint
Publicado:	2026
Materias:	Artificial Intelligence Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

MetaSC: Test-Time Safety Specification Optimization for Language Models
por: Gallego, Víctor
Publicado: (2025)

Configurable Safety Tuning of Language Models with Synthetic Preference Data
por: Gallego, Victor
Publicado: (2024)

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement
por: Gallego, Víctor
Publicado: (2025)

Configurable Preference Tuning with Rubric-Guided Synthetic Data
por: Gallego, Víctor
Publicado: (2025)

Merging Improves Self-Critique Against Jailbreak Attacks
por: Gallego, Victor
Publicado: (2024)

Signals: Trajectory Sampling and Triage for Agentic Interactions
por: Chen, Shuguang, et al.
Publicado: (2026)

LLM Can be a Dangerous Persuader: Empirical Study of Persuasion Safety in Large Language Models
por: Liu, Minqian, et al.
Publicado: (2025)

ARM: Discovering Agentic Reasoning Modules for Generalizable Multi-Agent Systems
por: Yao, Bohan, et al.
Publicado: (2025)

RAGalyst: Automated Human-Aligned Agentic Evaluation for Domain-Specific RAG
por: Gao, Joshua, et al.
Publicado: (2025)

LangFIR: Discovering Sparse Language-Specific Features from Monolingual Data for Language Steering
por: Wong, Sing Hieng, et al.
Publicado: (2026)

Soteria: Language-Specific Functional Parameter Steering for Multilingual Safety Alignment
por: Banerjee, Somnath, et al.
Publicado: (2025)

Verus-SpecGym: An Agentic Environment for Evaluating Specification Autoformalization
por: Agarwal, Anmol, et al.
Publicado: (2026)

Agentic Adversarial QA for Improving Domain-Specific LLMs
por: Grari, Vincent, et al.
Publicado: (2026)

Uncovering the Potential Risks in Unlearning: Danger of English-only Unlearning in Multilingual LLMs
por: Hwang, Kyomin, et al.
Publicado: (2025)

Discover and Prove: An Open-source Agentic Framework for Hard Mode Automated Theorem Proving in Lean 4
por: Liu, Chengwu, et al.
Publicado: (2026)

SWE-AGI: Benchmarking Specification-Driven Software Construction with MoonBit in the Era of Autonomous Agents
por: Zhang, Zhirui, et al.
Publicado: (2026)

Bi-Mamba: Towards Accurate 1-Bit State Space Models
por: Tang, Shengkun, et al.
Publicado: (2024)

Agentic Reward Modeling: Integrating Human Preferences with Verifiable Correctness Signals for Reliable Reward Systems
por: Peng, Hao, et al.
Publicado: (2025)

Safe in the Future, Dangerous in the Past: Dissecting Temporal and Linguistic Vulnerabilities in LLMs
por: Said, Muhammad Abdullahi, et al.
Publicado: (2025)

Output Length Effect on DeepSeek-R1's Safety in Forced Thinking
por: Li, Xuying, et al.
Publicado: (2025)

Using LLMs to Discover Legal Factors
por: Gray, Morgan, et al.
Publicado: (2024)

Subjective $\textit{Isms}$? On the Danger of Conflating Hate and Offence in Abusive Language Detection
por: Curry, Amanda Cercas, et al.
Publicado: (2024)

An Agentic AI Framework for Training General Practitioner Student Skills
por: De Marez, Victor, et al.
Publicado: (2025)

Discovering Significant Topics from Legal Decisions with Selective Inference
por: Soh, Jerrold
Publicado: (2024)

LittleBit: Ultra Low-Bit Quantization via Latent Factorization
por: Lee, Banseok, et al.
Publicado: (2025)

Budget-Aware Agentic Routing via Boundary-Guided Training
por: Zhang, Caiqi, et al.
Publicado: (2026)

AgentAlign: Navigating Safety Alignment in the Shift from Informative to Agentic Large Language Models
por: Zhang, Jinchuan, et al.
Publicado: (2025)

Resisting Correction: How RLHF Makes Language Models Ignore External Safety Signals in Natural Conversation
por: Cataneo, Felipe Biava
Publicado: (2025)

NILC: Discovering New Intents with LLM-assisted Clustering
por: Wang, Hongtao, et al.
Publicado: (2025)

SafeLLM: Domain-Specific Safety Monitoring for Large Language Models: A Case Study of Offshore Wind Maintenance
por: Walker, Connor, et al.
Publicado: (2024)

Safety Alignment Can Be Not Superficial With Explicit Safety Signals
por: Li, Jianwei, et al.
Publicado: (2025)

Discovering Process-Outcome Credit in Multi-Step LLM Reasoning
por: Wang, Xiangwei, et al.
Publicado: (2026)

Precision Knowledge Editing: Enhancing Safety in Large Language Models
por: Li, Xuying, et al.
Publicado: (2024)

LLM Reasoning as Trajectories: Step-Specific Representation Geometry and Correctness Signals
por: Sun, Lihao, et al.
Publicado: (2026)

Agentic Confidence Calibration
por: Zhang, Jiaxin, et al.
Publicado: (2026)

Agentic Uncertainty Quantification
por: Zhang, Jiaxin, et al.
Publicado: (2026)

AgenticSum: An Agentic Inference-Time Framework for Faithful Clinical Text Summarization
por: Piya, Fahmida Liza, et al.
Publicado: (2026)

Agentic Reasoning: A Streamlined Framework for Enhancing LLM Reasoning with Agentic Tools
por: Wu, Junde, et al.
Publicado: (2025)

Can Reinforcement Learning Unlock the Hidden Dangers in Aligned Large Language Models?
por: Karkevandi, Mohammad Bahrami, et al.
Publicado: (2024)

R1-ACT: Efficient Reasoning Model Safety Alignment by Activating Safety Knowledge
por: In, Yeonjun, et al.
Publicado: (2025)