Image de couverture de livre

Why Does Agentic Safety Fail to Generalize Across Tasks?

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Slutzky, Yonatan, Alexander, Yotam, Slor, Tomer, Nagel, Yoav, Cohen, Nadav
Format:	Preprint
Publié:	2026
Sujets:	Machine Learning
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

The Implicit Bias of Structured State Space Models Can Be Poisoned With Clean Labels
par: Slutzky, Yonatan, et autres
Publié: (2024)

Do Neural Networks Need Gradient Descent to Generalize? A Theoretical Study
par: Alexander, Yotam, et autres
Publié: (2025)

Outcome-Based RL Provably Leads Transformers to Reason, but Only With the Right Data
par: Ran-Milo, Yuval, et autres
Publié: (2026)

Deep Learning for Optical Misalignment Diagnostics in Multi-Lens Imaging Systems
par: Slor, Tomer, et autres
Publié: (2025)

What Makes Data Suitable for a Locally Connected Neural Network? A Necessary and Sufficient Condition Based on Quantum Entanglement
par: Alexander, Yotam, et autres
Publié: (2023)

On the Expressive Power of Sparse Geometric MPNNs
par: Sverdlov, Yonatan, et autres
Publié: (2024)

Implicit Bias of Policy Gradient in Linear Quadratic Control: Extrapolation to Unseen Initial States
par: Razin, Noam, et autres
Publié: (2024)

NnD: Diffusion-based Generation of Physically-Nonnegative Objects
par: Torem, Nadav, et autres
Publié: (2025)

Cluster Frequency Conformal Prediction for Local Coverage
par: Lavi, Tomer, et autres
Publié: (2026)

Trajectory-Based Difficulty Scoring for Reliable Learning on Tabular Data
par: Lavi, Tomer, et autres
Publié: (2026)

Clustered Calibration: Representation-Aware Probability Calibration via Learned Subpopulations
par: Lavi, Tomer, et autres
Publié: (2025)

Lecture Notes on Linear Neural Networks: A Tale of Optimization and Generalization in Deep Learning
par: Cohen, Nadav, et autres
Publié: (2024)

When and How to Canonize: A Generalization Perspective
par: Sverdlov, Yonatan, et autres
Publié: (2026)

Revisiting Multi-Permutation Equivariance through the Lens of Irreducible Representations
par: Sverdlov, Yonatan, et autres
Publié: (2024)

Why Domain Generalization Fail? A View of Necessity and Sufficiency
par: Vuong, Long-Tung, et autres
Publié: (2025)

Mamba Knockout for Unraveling Factual Information Flow
par: Endy, Nir, et autres
Publié: (2025)

Adaptation Odyssey in LLMs: Why Does Additional Pretraining Sometimes Fail to Improve?
par: Öncel, Fırat, et autres
Publié: (2024)

Generative Myopia: Why Diffusion Models Fail at Structure
par: Siami, Milad
Publié: (2025)

BXRL: Behavior-Explainable Reinforcement Learning
par: Rachum, Ram, et autres
Publié: (2026)

Monotone and Separable Set Functions: Characterizations and Neural Models
par: Sarangi, Soutrik, et autres
Publié: (2025)

FSW-GNN: A Bi-Lipschitz WL-Equivalent Graph Neural Network
par: Sverdlov, Yonatan, et autres
Publié: (2024)

Short-Range Oversquashing
par: Mishayev, Yaaqov, et autres
Publié: (2025)

Why Model Selection Fails in Time Series Forecasting: An Empirical Study of Instability Across Data Regimes
par: Akinci, Tahir Cetin, et autres
Publié: (2026)

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

Inertial Navigation Meets Deep Learning: A Survey of Current Trends and Future Directions
par: Cohen, Nadav, et autres
Publié: (2023)

Tools Fail: Detecting Silent Errors in Faulty Tools
par: Sun, Jimin, et autres
Publié: (2024)

Failing to Explore: Language Models on Interactive Tasks
par: JafariRaviz, Mahdi, et autres
Publié: (2026)

In-context Learning and Gradient Descent Revisited
par: Deutch, Gilad, et autres
Publié: (2023)

Why Attention Fails: The Degeneration of Transformers into MLPs in Time Series Forecasting
par: Liang, Zida, et autres
Publié: (2025)

When Explanations Lie: Why Many Modified BP Attributions Fail
par: Sixt, Leon, et autres
Publié: (2019)

Conditional Balance: Improving Multi-Conditioning Trade-Offs in Image Generation
par: Cohen, Nadav Z., et autres
Publié: (2024)

Gradient Starvation in Binary-Reward GRPO: Why Group-Mean Centering Fails and Why the Simplest Fix Works
par: Nie, Wenhua, et autres
Publié: (2026)

Why Do More Experts Fail? A Theoretical Analysis of Model Merging
par: Wang, Zijing, et autres
Publié: (2025)

Why Smooth Stability Assumptions Fail for ReLU Learning
par: Katende, Ronald
Publié: (2025)

Why Do Transformers Fail to Forecast Time Series In-Context?
par: Zhou, Yufa, et autres
Publié: (2025)

SINR-Aware Deep Reinforcement Learning for Distributed Dynamic Channel Allocation in Cognitive Interference Networks
par: Cohen, Yaniv, et autres
Publié: (2024)

Improved Distribution Estimation in $\ell_\infty$
par: Cohen, Doron, et autres
Publié: (2026)

Consensus is Not Verification: Why Crowd Wisdom Strategies Fail for LLM Truthfulness
par: Denisov-Blanch, Yegor, et autres
Publié: (2026)

Why Low-Precision Transformer Training Fails: An Analysis on Flash Attention
par: Qiu, Haiquan, et autres
Publié: (2025)

Substance Beats Style: Why Beginning Students Fail to Code with LLMs
par: Lucchetti, Francesca, et autres
Publié: (2024)