Search: “safety”

Revisiting Topological Graphs for Macro Action based Closed-loop Reinforcement Learning of Vision Language Navigation in Continuous Environment

arXiv:2609.03906v1 Announce Type: new Abstract: Vision-Language Navigation in Continuous Environments (VLN-CE) requires an agent to follow natural language instructions through unseen environments. Ex

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

RoboTok: An Internet-Scale Data Engine for Human Demonstration Retrieval and Dexterous Manipulation Learning

DGX agent

arXiv:2609.03199v1 Announce Type: new Abstract: Robot learning increasingly depends on broad and diverse demonstrations, yet collecting robot data remains expensive and poorly suited to covering the l

safetyarxiv-cs-cv

4 Sept 2026

SafeRestore: Detector-Relative Risk Certificates for Selective Industrial Image Restoration

arXiv:2609.03475v1 Announce Type: new Abstract: Industrial inspection pipelines often restore a measured image before a detector acts on it, yet restoration can suppress detector-supported defect stru

ScoreMix: Synthetic Data Generation by Score Composition in Diffusion Models Improves Recognition

arXiv:2506.10226v3 Announce Type: replace-cross Abstract: Synthetic data generation is increasingly used in machine learning for training and data augmentation. Yet, current strategies often rely on e

Select, Compress, Reinvest: A Controlled Study of Visual-Token Allocation in Long-Video MLLMs

arXiv:2609.03820v1 Announce Type: cross Abstract: Long-video language models cannot look at every frame: an hour sampled once per second is 3,600 images, and a system keeps only a small fixed slice of

SignSeek: Learning Transferable Representations for Sign Dictionary Retrieval

arXiv:2609.03695v1 Announce Type: new Abstract: Sign language dictionaries are essential resources for sign language learners, yet automatically retrieving a sign from a dictionary, given only a query

sounds like the government testing standard fail to adequately factor in monitorability. that’s bad.

sounds like the government testing standard fail to adequately factor in monitorability. that’s bad. OpenAI announced its latest model, Astra today. When asked during a briefing if they followed the n

safetygary-marcus--x

Spectral Hierarchy of the Cosmic Web

arXiv:2603.15834v2 Announce Type: replace-cross Abstract: We introduce a spectral hierarchy of cosmic-web classifications obtained by applying simple scale-weighting kernels to the density field befor

SWIM: Student Writing Simulation via Proficiency-Conditioned Generation

arXiv:2609.03215v1 Announce Type: new Abstract: Writing proficiency manifests in how students develop content, organize ideas, choose words, and use language. Despite growing interest in LLM-based stu

The Attention Triangle in Audio-Video Models

arXiv:2609.03586v1 Announce Type: new Abstract: Audio-video diffusion models rely on cross-modal attention to coordinate text, sound, and visual content, yet this same mechanism can introduce subtle a

The Dice Roll Method: A Standardized Protocol for Repeated-Query Auditing of Large Language Model Brand Recommendations

arXiv:2609.04047v1 Announce Type: cross Abstract: Background: Researchers increasingly use repeated identical prompts to audit stochastic variation in large language model (LLM) brand recommendations,

⚠️ This is really scary. Proxies for OpenAI are laying the groundwork for saying the rest of us should suck up rogue AI risk and quit our bi…

⚠️ This is really scary. Proxies for OpenAI are laying the groundwork for saying the rest of us should suck up rogue AI risk and quit our bitching. Because they know they are on the verge of losing co

safetygary-marcus--x

TIGPO: Temporal Instance-Graph Policy Optimization for Long-Horizon LLM Agents

arXiv:2609.03383v1 Announce Type: new Abstract: Graph-based policy optimization improves credit assignment for long-horizon LLM agents by organizing rollout trajectories into state-transition graphs.

Toward an~Integrated Cognitive--Ergonomic Architecture for~Human--Machine Interaction: Combining Cognitive Models with~Human Factors Ergonomics

arXiv:2609.03704v1 Announce Type: new Abstract: This paper presents an integrated approach to modeling human competencies by combining the theoretical foundations of cognitive architectures with princ

Toward Unified Robot Learning: Bridging Representation, Vision-Language-Action, and World Models

arXiv:2609.03927v1 Announce Type: new Abstract: For robots to operate reliably in real-world environments, they need to perceive their surroundings, act, and reason about the consequences of those act

TRaIL-Odom: Tightly Coupled Continuous Time Radar-IMU-LiDAR Odometry with Adaptive Doppler Weighting

arXiv:2609.03561v1 Announce Type: new Abstract: Existing radar-LiDAR fusion methods rely on fixed residual weights, even though the informativeness of radar Doppler and LiDAR geometry is scan- and dir

Two-Stage Reinforcement Learning for Sound and Adversarial Test Generation in Code LLMs

arXiv:2609.03955v1 Announce Type: new Abstract: Reinforcement learning (RL) has substantially advanced code generation with large language models (LLMs) through executable feedback. The feedback for c

Unfold The World: Factorize 4D Properties in Reinforcing Spatial Reasoning

arXiv:2609.03729v1 Announce Type: new Abstract: Despite the remarkable prowess of Vision-Language Models (VLMs) in general multimodal tasks, they remain fundamentally ``flat'' when reasoning about the

User Perceptions vs. Proxy LLM Judges: Privacy and Helpfulness in LLM Responses to Privacy-Sensitive Scenarios

arXiv:2510.20721v4 Announce Type: replace-cross Abstract: Large language models (LLMs) are rapidly being adopted for tasks like drafting emails, summarizing meetings, and answering health questions. I

WISE: World-model-guided Imagination Scheduling for Efficient Post-training of Vision-Language-Action Models

arXiv:2609.03681v1 Announce Type: new Abstract: Post-training VLA policies typically rely on supervised fine-tuning with costly expert demonstrations or reinforcement learning with expensive and poten

A Unified Rate-Distortion Perspective on Vector, Product, and Scalar Quantization

arXiv:2609.02107v1 Announce Type: cross Abstract: Discrete visual tokenization, predominantly driven by vector, scalar, and product quantization, lacks a unified conceptual framework for understanding

Accurate in space, unreliable in time: how LLMs represent national cultural change

arXiv:2609.01902v1 Announce Type: cross Abstract: Assessments of cultural alignment have become an important part of the development and improvement of large language models (LLMs). However, the major

Act More, Decide Less: Skill-Guided Adaptive Action Chunking for Long-Horizon LLM Agents

arXiv:2609.02042v1 Announce Type: new Abstract: Large language model (LLM) agents for long-horizon interactive tasks typically follow a ReAct-style protocol, issuing one primitive action per LLM round

Action abstractions for amortized sampling

arXiv:2410.15184v2 Announce Type: replace-cross Abstract: As trajectories sampled by policies used by reinforcement learning (RL) and generative flow networks (GFlowNets) grow longer, credit assignmen

AffectDelta: Beyond Emotion Labels for Image Editing

arXiv:2609.02616v1 Announce Type: new Abstract: Emotion-driven image editing aims to evoke a specified target emotion by modifying emotion-relevant visual cues in a source image, while preserving the

AlphaRAD: Grounded Zero-Shot Classification in Chest Radiology via alpha-Corrected Binary Cross Entropy and Factorized Latent Supervision

arXiv:2609.01757v1 Announce Type: new Abstract: Vision-Language Pretrained Models (VLPMs) offer a scalable path to open-vocabulary chest radiology understanding, yet two aspects remain underexplored:

Anyone who cared about the science of AI would find this interesting. But if you are just here for the religion, I guess you don’t?

Anyone who cared about the science of AI would find this interesting. But if you are just here for the religion, I guess you don’t? 🚨Breaking: The success of GPT-6 Astra on ARC-AGI-3 provides strong e

safetygary-marcus--x

Are Non-English Papers Reviewed Fairly? Language-of-Study Bias in NLP Peer Reviews

arXiv:2604.07119v2 Announce Type: replace Abstract: Peer review plays a central role in the NLP publication process, but is susceptible to various biases. Here, we study language-of-study (LoS) bias:

As AI security concerns rise, HiddenLayer raises 100M and Huskeys bags 27M

Artificial intelligence security startup HiddenLayer Inc. today announced a 100 million Series B round, and a separate 27 million Series A went to edge security company Huskeys Security Ltd. Austin-ba

safetysiliconangle

Balancing Frequencies and Pixels in Flow Matching

arXiv:2609.02748v1 Announce Type: new Abstract: Natural images follow a 1/f^2 spectral distribution: most signal energy lies in the low spatial frequencies, while the perceptually important structures

Beyond Modality Harmony: Orthogonal Purification and Topology-Guided MoE for Conflict-Aware Multimodal Recommendation

arXiv:2609.02152v1 Announce Type: cross Abstract: Multimodal Recommender Systems (MRSs) typically rely on a flawed 'modality harmony' assumption, presuming that multimodal features are inherently bene

Beyond Outcome Gaps: Process-Aware Fairness Diagnosis for LLM-based Multi-Agent Decision Systems

arXiv:2609.02092v1 Announce Type: new Abstract: LLM-based multi-agent systems (MAS) are increasingly considered for high-stakes decision-making, yet outcome-based fairness audits can miss where risks

Breadth Beats Depth: Improving GCG-Based Jailbreak Optimization with Breadth-Oriented Suffix Search

arXiv:2609.02172v1 Announce Type: new Abstract: Optimization-based jailbreak attacks such as Greedy Coordinate Gradient (GCG) achieve strong effectiveness and transferability by optimizing adversarial

Brilliant paper on long-horizon agents. They cut 78.9% of an agent's LLM calls while raising its success rate. Here is how: It turns out tha…

Brilliant paper on long-horizon agents. They cut 78.9% of an agent's LLM calls while raising its success rate. Here is how: It turns out that ReAct issues one primitive action per model round. That al

safetydair-ai--x

CA-OPD: Confidence-Aware On-Policy Distillation for Structured Visual Prediction

arXiv:2609.02401v1 Announce Type: new Abstract: Autoregressive vision language models unify heterogeneous perception tasks but are highly susceptible to compounding errors. On-policy distillation (OPD

CHisAgent: A Multi-Agent Framework for Event Taxonomy Construction in Ancient Chinese Cultural Systems

arXiv:2601.05520v2 Announce Type: replace Abstract: Despite strong performance on many tasks, large language models (LLMs) show limited ability in historical and cultural reasoning, particularly in no

CLASE: A Hybrid Method for Chinese Legalese Stylistic Evaluation

arXiv:2602.12639v2 Announce Type: replace Abstract: Legal text generated by large language models (LLMs) can usually achieve reasonable factual accuracy, but it frequently fails to adhere to the speci

Cliff: Learning Process Rewards from the First Mistake

arXiv:2609.02817v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for large language model (LLM) post-training, but its reliance

Competitive Market Behavior of LLMs

arXiv:2609.02580v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly deployed as economic agents, yet there is little evidence whether LLM agents are suited for participatin

Contrastive Explanations in Quantitative Bipolar Argumentation Frameworks

arXiv:2609.02399v1 Announce Type: new Abstract: Argumentation frameworks are useful tools for representing and reasoning with information in a variety of settings, e.g. in supplementing AI models as t

Debias-SparseGPT: Bias-Aware Pruning for Large Language Models

arXiv:2609.02496v1 Announce Type: new Abstract: Model compression techniques such as pruning and quantization facilitate the efficient deployment and acceleration of Large Language Models (LLMs). Howe

Deep Reinforcement Learning for Reach-Avoid-Stay Problems

arXiv:2410.02898v3 Announce Type: replace-cross Abstract: Reach-Avoid-Stay (RAS) tasks are essential in applications where systems must safely reach a target set and remain within it under all bounded

DMRL: Document-Mediated Reinforcement Learning for Skill Optimization in Advertising Recommendation

arXiv:2609.02170v1 Announce Type: new Abstract: Advertising recommendation requires continuously tuning complex system parameters while balancing commercial returns and user experience. Recent work ha

Do Better Imagined Rollouts Mean Better Robot Control? A Controlled Study of World-Model Evaluation Under Feedback

arXiv:2609.02811v1 Announce Type: new Abstract: Predictive models are increasingly used in robotics for state estimation, planning, control, and policy evaluation, yet they are often judged by open-lo

Do Cantonese-Adapted Language Models Better Predict Cantonese Reading? A Cross-Model Eye-Tracking Evaluation

arXiv:2609.02163v1 Announce Type: new Abstract: Information-theoretic measures derived from autoregressive language models are widely used to characterize the expectations that shape human reading, bu

Efficient Context-Limited Telescope Bibliography Classification for the WASP-2025 Shared Task Using SciBERT

arXiv:2609.01647v1 Announce Type: new Abstract: The creation of telescope bibliographies is a crucial part of assessing the scientific impact of observatories and ensuring reproducibility in astronomy

Eliciting ESG Preferences for Reinforcement Learning-Based Portfolio Optimization

arXiv:2609.02677v1 Announce Type: cross Abstract: Modern portfolio management increasingly demands a balance between traditional risk-adjusted returns and strict Environmental, Social, and Governance

Examining the Vulnerability of Multi-Agent Medical Systems to Human Interventions for Clinical Reasoning

arXiv:2609.02191v1 Announce Type: new Abstract: Human interventions at fault points can alter the diagnostic accuracy of multi-agent medical systems. We defined fault points as moments in AI agent con