safety

Identifiable Convex-Concave Regression via Sub-gradient Regularised Least Squares

arXiv:2506.18078v2 Announce Type: replace-cross Abstract: We propose a novel nonparametric regression method that models complex input-output relationships as the sum of convex and concave components.

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Inductive Feedback for Mixed-Policy Distillation

DGX agent

arXiv:2609.35390v1 Announce Type: new Abstract: Verbal feedback can identify errors and prescribe corrections, providing rich supervision for language-model post-training even when reliable programmat

safetyarxiv-cs-lg

29 Sept 2026

InfraVLA: Extending Vision-Language-Action Navigation with Infrastructure Cameras

arXiv:2609.33647v1 Announce Type: new Abstract: Many indoor environments in which robots operate, such as warehouses, offices, and hospitals, already have cameras installed. They observe parts of the

Instruct, Not Answer: Using Instruction Privileges in On-Policy Context Distillation

arXiv:2609.32201v1 Announce Type: cross Abstract: On-Policy Context Distillation (OPCD) has recently emerged as a powerful technique for transferring context to student models and for self-improvement

Integrity Detection and Characterization of Malicious Injections in RAVEN II

arXiv:2609.33951v1 Announce Type: new Abstract: The increasing adoption of robotic systems in surgery, together with the expanding range of procedures they can support and the growing level of autonom

👀 Interesting data from @tickerplus. Acc to this data, Anthropic's ARR seems to have flatlined around the time when the token index peaked.…

👀 Interesting data from @tickerplus. Acc to this data, Anthropic's ARR seems to have flatlined around the time when the token index peaked. Many people questioned (rightly) at the time the inferential

safetygary-marcus--x

Interference Beyond Geometry in Concept Extraction

arXiv:2609.35351v1 Announce Type: new Abstract: Interference is commonly treated as geometric overlap between learned features. We introduce effective interference, which combines feature geometry and

Intuition vectors

arXiv:2609.32619v1 Announce Type: new Abstract: Large self-supervised vision models learn representations that support scene segmentation and the semantic decomposition of physical objects. We ask whe

Is invariance all you need for algorithmic fairness? Removing demographic information can create new bias

arXiv:2609.32004v1 Announce Type: new Abstract: Encoded demographic information in internal model representations is a commonly assumed risk factor for algorithmic bias, with demographic representatio

Is Online Interaction Necessary for Recovery? A Minimalist Approach to Robust Planning via Perturbation

arXiv:2609.33049v1 Announce Type: new Abstract: Behavior cloning (BC) is vulnerable to covariate shift during closed-loop execution, where small prediction or execution errors can drive the robot towa

Just-In-Time Agent Memory with Runtime Agentic Research

arXiv:2609.34385v1 Announce Type: cross Abstract: Memory is critical for AI agents. Many existing agent-memory systems follow an Ahead-of-Time (AOT) design, constructing memory before a specific reque

🚨 JUST NOW: Elon Musk and Jensen Huang give an INSIDE LOOK at the big meeting between top AI firms and President Trump today, saying they'l…

🚨 JUST NOW: Elon Musk and Jensen Huang give an INSIDE LOOK at the big meeting between top AI firms and President Trump today, saying they'll now be 'grading each other's homework' 😆 @elonmusk: 'We sig

safetyelon-musk--x

K-OPSD: Verifiable On-Policy Self-Distillation for Post-Training Vision-Language Models on AEC Drawings

arXiv:2609.34082v1 Announce Type: cross Abstract: Interpreting architecture, engineering, and construction (AEC) drawings is hard for general Multimodal Large Language Models (MLLMs) and vision-langua

lapanda: A Matrix-Free Differentiable Solver for Nonconvex Constrained Optimization Layers

arXiv:2609.33630v1 Announce Type: cross Abstract: Differentiable optimization brings the structural guarantees of mathematical optimization to network pipelines, allowing them to be trained end-to-end

Last-Iterate Guarantees for Online Reinforcement Learning in Structured Constrained MDPs

arXiv:2609.32933v1 Announce Type: new Abstract: In safety-critical applications, deployment uses a single policy, whose performance and constraint satisfaction should hold directly rather than only fo

Leaky Students: Membership Inference against On-Policy Distillation

arXiv:2609.33136v1 Announce Type: new Abstract: On-policy distillation (OPD) trains a student to match a teacher's next-token distributions on student-generated trajectories. However, privileged infor

Learning beyond Site Bias for OOD Generalization in Brain Networks

arXiv:2605.06050v2 Announce Type: replace Abstract: Graph-based learning from functional magnetic resonance imaging (fMRI) has shown strong potential for brain network analysis. However, existing meth

Learning Geometrically-Grounded Amodal 3D Representations for View-Generalizable Robotic Manipulation

arXiv:2601.22988v2 Announce Type: replace Abstract: Real-world robotic manipulation requires visuomotor policies capable of robust 3D scene reasoning under varying camera viewpoints. While recent 3D-a

Learning High-Risk High-Precision Motion Control

arXiv:2609.34851v1 Announce Type: new Abstract: Deep reinforcement learning (DRL) algorithms for movement control are typically evaluated and benchmarked on sequential decision tasks where imprecise a

Learning Perturbation Robust Policies for LLM Agents with Stable Optimization

arXiv:2609.34064v1 Announce Type: new Abstract: Reinforcement learning (RL) has become an effective post-training paradigm for long-horizon large language model (LLM) agents. However, we find that the

Learning Transferable Sensor Models via Language-Informed Pretraining

arXiv:2603.11950v2 Announce Type: replace-cross Abstract: Multimodal language models have demonstrated strong semantic understanding and reasoning over physiological and behavioral signals captured fr

Learning What to Evaluate: Correlation-Aware Decoupling for Multiobjective Bayesian Optimization

arXiv:2609.32632v1 Announce Type: cross Abstract: Multiobjective Bayesian optimization (MOBO) with Gaussian process (GP) surrogates is a sample efficient approach to solving multiobjective optimizatio

Learning with Object-centric Representations of Tactile Interactive Perception for Robot Manipulation

arXiv:2609.33235v1 Announce Type: new Abstract: Implicit object properties that are difficult to directly infer from vision, such as material, container contents, or softness, can be revealed through

Less Sycophancy, Stronger Refusal? Lessons for AI Safety from Mechanistic Interpretability

arXiv:2609.35544v1 Announce Type: cross Abstract: Reliable refusal of harmful requests is essential to the safe deployment of language models. Because excessive eagerness to please users may undermine

Let CSP Be Your ANCHOR: Adaptive Crystal Search over Frozen Structure Priors

arXiv:2609.33407v1 Announce Type: cross Abstract: De novo crystal generation (DNG) models decide where to search in composition space and how to generate structures with one set of weights. We argue t

Levy-Driven Correspondence Estimation for Registration

arXiv:2609.32612v1 Announce Type: new Abstract: Finding reliable point correspondences is difficult when point clouds have low overlap or undergo non-rigid deformation. Iterative refinement can correc

LFPO: Likelihood-Free Policy Optimization for Masked Diffusion Models

arXiv:2603.01563v2 Announce Type: replace Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has achieved remarkable success in improving autoregressive models, especially in domains requ

LOD: Latent Objective Discovery in Heterogeneous Multi-Agent Reinforcement Learning

arXiv:2505.09756v2 Announce Type: replace Abstract: In multi-agent reinforcement learning (MARL) settings, agents are often heterogeneous with diverse intrinsic utility functions. In this study, we in

Long-Horizon Scaling: How Model Capabilities Shape the Returns to Computation

arXiv:2609.35236v1 Announce Type: new Abstract: Long-horizon agents improve solutions through sustained interaction, execution, and task feedback. Scaling studies relate performance to resources and c

Low-Confidence Remasking Traps Flexibility: Realizing Arbitrary-Order Potential for Diverse Rollouts in Diffusion LLMs

arXiv:2609.34509v1 Announce Type: new Abstract: Masked diffusion language models support arbitrary-order generation, suggesting a natural way to produce diverse outputs. However, recent work argues th

MaD-RL: Matching Distributions for Calibrating LLMs with Reinforcement Learning

arXiv:2609.31644v1 Announce Type: new Abstract: Reinforcement learning (RL) is widely used in language-model post-training to maximize rewards assigned to individual model outputs, such as scores from

MaPP: A Unified Marginalized Posterior-Predictive Framework for Data-Efficient RLVR

arXiv:2609.34990v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models but incurs substantial costs from rol

Mask2Restore: Self-Supervised Ultrasound Despeckling via Inpainting

arXiv:2609.32844v1 Announce Type: cross Abstract: Medical ultrasound (US) is inherently degraded by speckle, a granular interference pattern that is often treated as a complex form of noise in image r

MegaGraph: Towards Efficient Training of Large-Scale Graph Transformers with Automated Hybrid Parallelism

arXiv:2609.34420v1 Announce Type: new Abstract: Graph Transformers (GTs) offer superior representation capabilities by overcoming the depth limitations and over-smoothing issues of traditional Graph N

MinkowskiPE: Minkowski Positional Encoding for Spatiotemporal Perception

arXiv:2609.33804v1 Announce Type: cross Abstract: Modeling spatiotemporal coupling is a key challenge in building physical intelligence across scales, from microscopic to macroscopic. Existing models

MM-OPD: Towards One More Bottleneck Between Perception and Reasoning

arXiv:2609.32690v1 Announce Type: new Abstract: Recent multimodal large language models (MLLMs) advance visual reasoning by strengthening both perception and reasoning, implicitly assuming a process t

mmHRI: Towards Privacy-Preserving Human-Robot Interaction with Millimeter-Wave Radar

arXiv:2609.34220v1 Announce Type: cross Abstract: Assistive robots increasingly operate in many human-centered environments and perform various human-robot interaction (HRI) tasks, such as object deli

Model-Free Output Feedback Stabilization via Policy Gradient Methods

arXiv:2601.19284v3 Announce Type: replace-cross Abstract: Stabilizing a dynamical system is a fundamental problem that serves as a cornerstone for many complex tasks in the field of control systems. T

Model-Informed Safe Reinforcement Learning for Bipedal Locomotion via Step-to-Step Prediction

arXiv:2609.34486v1 Announce Type: new Abstract: Humanoid robots promise versatile mobility in cluttered, human-centric environments, but real deployment demands principled safety. Classical model-base

Modelling non-linear aeroelastic loads in long-span bridges with extreme learning machines

arXiv:2609.33274v1 Announce Type: cross Abstract: Accurate modelling of aerodynamic loads is essential for predicting instabilities and ensuring the safety of long-span bridges. A methodology is intro

MonoEgo: Monocular Metric Egocentric Demonstration Capture with Passive Wrist Constellations and Sparse Workstation Anchors

arXiv:2609.34512v1 Announce Type: new Abstract: Image-aligned metric demonstrations often require dedicated tracking hardware and synchronization across devices. We present MonoEgo, a capture system t

Multi-Scale Semantic Mapping in Urban Environments via Observation Calibration and Policy Dependence Regularization

arXiv:2609.34833v1 Announce Type: new Abstract: Semantic mapping is fundamental to embodied navigation, yet existing methods are developed for indoor environments, where objects exhibit relatively lim

Narrow Multimodal Fine-Tuning Can Induce Emergent Misalignment

arXiv:2609.35291v1 Announce Type: cross Abstract: Modern AI models are aligned through post-training to adapt them to downstream tasks. Recent work shows that fine-tuning language models on narrow tas

Natural State-Prediction Accuracy can Hide Weak Controlled Responsiveness in VLA Readouts

arXiv:2609.34684v1 Announce Type: cross Abstract: Accurately decoding object states from the internal representations of vision-language-action (VLA) models does not establish that the predictions res

No Free Efficiency: Revisiting the Trade-off Between Training Efficiency and Model Vulnerability

arXiv:2609.33898v1 Announce Type: new Abstract: Training efficiency has become the central driver of recent progress in foundation models. To overcome the massive computational and data requirements o

Not All Errors Matter: Decision-Relevant Prediction Error Predicts Planning Quality

arXiv:2609.32322v1 Announce Type: new Abstract: World models are typically trained and evaluated by prediction error, assuming that more accurate predictions lead to better decisions. We show that thi

On-Policy Attention Linearization

arXiv:2609.31947v1 Announce Type: new Abstract: Hybrid transformer architectures that replace most softmax attention layers with linear attention offer transformer-level quality at a fraction of the m

On the Interaction of Compressibility and Adversarial Robustness

arXiv:2507.17725v3 Announce Type: replace-cross Abstract: As demands for resource efficiency and safety in modern neural networks intensify, substantial research effort has gone into model compression